| Message ID | 1493340847-25720-1-git-send-email-kdinh@apm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 04/27/2017 08:54 PM, Khuong Dinh wrote: > From: Khuong Dinh <kdinh@apm.com> > > This patch makes pci-xgene-msi driver ACPI-aware and provides > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > > Signed-off-by: Khuong Dinh <kdinh@apm.com> > Signed-off-by: Duc Dang <dhdang@apm.com> > Acked-by: Marc Zyngier <marc.zyngier@arm.com> Thanks. Currently testing. Will review and followup shortly. Jon.
On 28/04/17 01:54, Khuong Dinh wrote: > From: Khuong Dinh <kdinh@apm.com> > > This patch makes pci-xgene-msi driver ACPI-aware and provides > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > > Signed-off-by: Khuong Dinh <kdinh@apm.com> > Signed-off-by: Duc Dang <dhdang@apm.com> > Acked-by: Marc Zyngier <marc.zyngier@arm.com> > --- > v2: > - Verify with BIOS version 3.06.25 and 3.07.09 > v1: > - Initial version > --- > drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- > 1 files changed, 32 insertions(+), 3 deletions(-) > > diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c > index f1b633b..00aaa3d 100644 > --- a/drivers/pci/host/pci-xgene-msi.c > +++ b/drivers/pci/host/pci-xgene-msi.c > @@ -24,6 +24,7 @@ > #include <linux/pci.h> > #include <linux/platform_device.h> > #include <linux/of_pci.h> > +#include <linux/acpi.h> > > #define MSI_IR0 0x000000 > #define MSI_INT0 0x800000 > @@ -39,7 +40,7 @@ struct xgene_msi_group { > }; > > struct xgene_msi { > - struct device_node *node; > + struct fwnode_handle *fwnode; > struct irq_domain *inner_domain; > struct irq_domain *msi_domain; > u64 msi_addr; > @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, > .free = xgene_irq_domain_free, > }; > > +#ifdef CONFIG_ACPI > +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) > +{ > + return xgene_msi_ctrl.fwnode; > +} > +#endif > + > static int xgene_allocate_domains(struct xgene_msi *msi) > { > msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, > @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > if (!msi->inner_domain) > return -ENOMEM; > > - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), > + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, > &xgene_msi_domain_info, > msi->inner_domain); > > @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > return -ENOMEM; > } > > +#ifdef CONFIG_ACPI > + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); > +#endif > return 0; > } > > @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) > {}, > }; > > +#ifdef CONFIG_ACPI > +static const struct acpi_device_id xgene_msi_acpi_ids[] = { > + {"APMC0D0E", 0}, > + { }, > +}; > +#endif > + > static int xgene_msi_probe(struct platform_device *pdev) > { > struct resource *res; > @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) > goto error; > } > xgene_msi->msi_addr = res->start; > - xgene_msi->node = pdev->dev.of_node; > + > + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); > + if (!xgene_msi->fwnode) { > + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); Please provide something other than NULL, such as the base address if the device. That's quite useful for debugging. > + if (!xgene_msi->fwnode) { > + dev_err(&pdev->dev, "Failed to create fwnode\n"); > + rc = ENOMEM; > + goto error; > + } > + } > + > xgene_msi->num_cpus = num_possible_cpus(); > > rc = xgene_msi_init_allocator(xgene_msi); > @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) > .driver = { > .name = "xgene-msi", > .of_match_table = xgene_msi_match_table, > + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), > }, > .probe = xgene_msi_probe, > .remove = xgene_msi_remove, > The code is trivial, but relies on the MSI controller to probe before the PCI bus. What enforces this? How is it making sure that this is not going to break in the next kernel release? As far as I can tell, there is no explicit dependency between the two, making it the whole thing extremely fragile. Thanks, M.
On 04/28/2017 01:11 AM, Jon Masters wrote: > On 04/27/2017 08:54 PM, Khuong Dinh wrote: >> From: Khuong Dinh <kdinh@apm.com> >> >> This patch makes pci-xgene-msi driver ACPI-aware and provides >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> >> Signed-off-by: Duc Dang <dhdang@apm.com> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> > > Thanks. Currently testing. Will review and followup shortly. Attaching boot logs before and after. I tested with an Intel e1000. Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI interrupts working. Please followup with answers to Marc's question about how you propose to guarantee boottime ordering of binding. I urgently require that AppliedMicro complete upstreaming of this. It should have been upstream a year ago. You're not the only ones with patches like this left around - I'm coming for everyone else, too. Tested-by: Jon Masters <jcm@jonmasters.org> [ 0.000000] Booting Linux on physical CPU 0x0 [ 0.000000] Linux version 4.11.0-rc8.xgene_msix1+ (root@mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #1 SMP Fri Apr 28 04:44:20 EDT 2017 [ 0.000000] Boot CPU: AArch64 Processor [500f0001] [ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '') [ 0.000000] bootconsole [uart8250] enabled [ 0.000000] efi: Getting EFI parameters from FDT: [ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24 [ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18 [ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78. [ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000 [ 0.000000] ACPI: Early table checksum verification disabled [ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM ) [ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013) [ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724) [ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724) [ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724) [ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724) [ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724) [ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724) [ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724) [ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724) [ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724) [ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013) [ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013) [ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013) [ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013) [ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200 [ 0.000000] ACPI: NUMA: Failed to initialise from firmware [ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff] [ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0 [ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff] [ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff] [ 0.000000] Zone ranges: [ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff] [ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff] [ 0.000000] Movable zone start for each node [ 0.000000] Early memory node ranges [ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff] [ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff] [ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff] [ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff] [ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff] [ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff] [ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff] [ 0.000000] On node 0 totalpages: 262144 [ 0.000000] DMA zone: 64 pages used for memmap [ 0.000000] DMA zone: 0 pages reserved [ 0.000000] DMA zone: 65536 pages, LIFO batch:1 [ 0.000000] Normal zone: 192 pages used for memmap [ 0.000000] Normal zone: 196608 pages, LIFO batch:1 [ 0.000000] psci: is not implemented in ACPI. [ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608 [ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536 [ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7 [ 0.000000] Detected PIPT I-cache on CPU0 [ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888 [ 0.000000] Policy zone: Normal [ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix1+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200 [ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes) [ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff] [ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved) [ 0.000000] Virtual kernel memory layout: [ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB) [ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB) [ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB) [ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB) [ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB) [ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB) [ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB) [ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB) [ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB) [ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum) [ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual) [ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB) [ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1 [ 0.000000] Hierarchical RCU implementation. [ 0.000000] Build-time adjustment of leaf fanout to 64. [ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8. [ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8 [ 0.000000] NR_IRQS:64 nr_irqs:64 0 [ 0.000000] GIC: Using split EOI/Deactivate mode [ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up [ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys). [ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns [ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns [ 0.008382] Console: colour dummy device 80x25 [ 0.013031] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000) [ 0.023774] pid_max: default: 32768 minimum: 301 [ 0.028553] ACPI: Core revision 20170119 [ 0.036208] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543) [ 0.049249] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790) [ 0.057204] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186) [ 0.063972] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543) [ 0.077125] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790) [ 0.084966] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186) [ 0.092086] ACPI: 4 ACPI AML tables successfully acquired and loaded [ 0.098656] Security Framework initialized [ 0.102953] SELinux: Initializing. [ 0.106548] SELinux: Starting in permissive mode [ 0.106747] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes) [ 0.116984] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes) [ 0.125658] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes) [ 0.132758] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes) [ 0.140711] ftrace: allocating 26814 entries in 7 pages [ 0.205140] ASID allocator initialised with 65536 entries [ 0.211274] Remapping and enabling EFI services. [ 0.216029] EFI remap 0x0000000010510000 => 0000000020000000 [ 0.222058] EFI remap 0x0000000010548000 => 0000000020018000 [ 0.228081] EFI remap 0x0000000017000000 => 0000000020020000 [ 0.234066] EFI remap 0x000000001c025000 => 0000000020035000 [ 0.240084] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware? [ 0.240089] ------------[ cut here ]------------ [ 0.252134] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104 [ 0.261010] Modules linked in: [ 0.265699] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix1+ #1 [ 0.273313] Hardware name: APM X-Gene Mustang board (DT) [ 0.278750] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000 [ 0.284827] PC is at efi_create_mapping+0x4c/0x104 [ 0.289772] LR is at efi_create_mapping+0x4c/0x104 [ 0.294664] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045 [ 0.302261] sp : ffff8003c1c03d20 [ 0.305710] x29: ffff8003c1c03d20 x28: 0000000000000000 [ 0.311183] x27: ffff000008c53000 x26: ffff000008d429c8 [ 0.316639] x25: ffff000008d42000 x24: 0000000000000000 [ 0.322102] x23: ffff000008a4c8f0 x22: ffff000008c60000 [ 0.327600] x21: ffff000008c53000 x20: ffff000008d429c8 [ 0.333038] x19: ffff8003f96ab348 x18: 000000003eca74e7 [ 0.338485] x17: 0000000000000000 x16: 0000000000000000 [ 0.343992] x15: ffff000008c53000 x14: 000000004e2b17ac [ 0.349464] x13: ffff0000089ca500 x12: 0000000000040b00 [ 0.354988] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0 [ 0.360435] x9 : 0000000000000073 x8 : 6620796767756220 [ 0.365898] x7 : 2d2d20424b203436 x6 : ffff000009274718 [ 0.371362] x5 : ffff8003ffdf7530 x4 : 0000000000000000 [ 0.376834] x3 : 0000000000000000 x2 : 0000000000040b00 [ 0.382316] x1 : 0000000000040b00 x0 : 0000000000000040 [ 0.389363] ---[ end trace 4a6935def63e1e16 ]--- [ 0.394126] Call trace: [ 0.396626] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60) [ 0.403246] 3b20: ffff8003f96ab348 0001000000000000 [ 0.411302] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8 [ 0.419331] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20 [ 0.427377] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8 [ 0.435449] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038 [ 0.443487] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8 [ 0.451585] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00 [ 0.459666] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530 [ 0.467729] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073 [ 0.475784] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500 [ 0.483840] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104 [ 0.489858] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238 [ 0.496817] [<ffff000008083698>] do_one_initcall+0x54/0x15c [ 0.502634] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c [ 0.508816] [<ffff0000087ae430>] kernel_init+0x18/0x108 [ 0.514202] [<ffff0000080833c0>] ret_from_fork+0x10/0x50 [ 0.519621] EFI remap 0x00000043fa7df000 => 000000002004f000 [ 0.525659] EFI remap 0x00000043fa82e000 => 000000002006e000 [ 0.531673] EFI remap 0x00000043ff9d6000 => 0000000020266000 [ 0.537700] EFI remap 0x00000043ff9ea000 => 000000002027a000 [ 0.543899] smp: Bringing up secondary CPUs ... [ 0.548738] Detected PIPT I-cache on CPU1 [ 0.548766] CPU1: Booted secondary processor [500f0001] [ 0.548987] Detected PIPT I-cache on CPU2 [ 0.549008] CPU2: Booted secondary processor [500f0001] [ 0.549225] Detected PIPT I-cache on CPU3 [ 0.549239] CPU3: Booted secondary processor [500f0001] [ 0.549432] Detected PIPT I-cache on CPU4 [ 0.549452] CPU4: Booted secondary processor [500f0001] [ 0.549653] Detected PIPT I-cache on CPU5 [ 0.549667] CPU5: Booted secondary processor [500f0001] [ 0.549861] Detected PIPT I-cache on CPU6 [ 0.549882] CPU6: Booted secondary processor [500f0001] [ 0.550086] Detected PIPT I-cache on CPU7 [ 0.550100] CPU7: Booted secondary processor [500f0001] [ 0.550142] smp: Brought up 1 node, 8 CPUs [ 0.621892] SMP: Total of 8 processors activated. [ 0.626806] CPU features: detected feature: 32-bit EL0 Support [ 0.632944] CPU: All CPU(s) started at EL2 [ 0.637984] devtmpfs: initialized [ 0.641749] SMBIOS 3.0.0 present. [ 0.645224] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016 [ 0.654642] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns [ 0.664967] futex hash table entries: 2048 (order: 2, 262144 bytes) [ 0.671981] pinctrl core: initialized pinctrl subsystem [ 0.677705] NET: Registered protocol family 16 [ 0.682714] cpuidle: using governor menu [ 0.686960] Detected 8 PCC Subspaces [ 0.690784] Registering PCC driver as Mailbox controller [ 0.696410] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000) [ 0.704152] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers. [ 0.713765] DMA: preallocated 256 KiB pool for atomic allocations [ 0.720154] ACPI: bus type PCI registered [ 0.724382] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5 [ 0.731202] Serial: AMBA PL011 UART driver [ 0.737465] HugeTLB registered 512 MB page size, pre-allocated 0 pages [ 0.744898] ACPI: Added _OSI(Module Device) [ 0.749274] ACPI: Added _OSI(Processor Device) [ 0.753882] ACPI: Added _OSI(3.0 _SCP Extensions) [ 0.758804] ACPI: Added _OSI(Processor Aggregator Device) [ 0.773904] ACPI: Interpreter enabled [ 0.777753] ACPI: Using GIC for interrupt routing [ 0.782685] ACPI: MCFG table detected, 1 entries [ 0.787684] HEST: Table parsing has been initialized. [ 0.795411] ACPI: Power Resource [SCVR] (on) [ 0.830232] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff]) [ 0.836749] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI] [ 0.845539] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability] [ 0.853915] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops [ 0.866664] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace [ 0.877881] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] [ 0.886116] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window] [ 0.893327] PCI host bridge to bus 0000:00 [ 0.897628] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff]) [ 0.908283] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff]) [ 0.919920] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window] [ 0.928114] pci_bus 0000:00: root bus resource [bus 00-ff] [ 0.933821] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400 [ 0.933899] pci 0000:00:00.0: supports D1 D2 [ 0.934059] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000 [ 0.934092] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff] [ 0.934108] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff] [ 0.934125] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f] [ 0.934141] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff] [ 0.934186] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref] [ 0.934326] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold [ 0.963877] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff] [ 0.971433] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff] [ 0.977908] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff] [ 0.985417] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref] [ 0.993304] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff] [ 1.000812] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff] [ 1.008298] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f] [ 1.014701] pci 0000:00:00.0: PCI bridge to [bus 01] [ 1.019893] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff] [ 1.026276] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff] [ 1.034419] vgaarb: loaded [ 1.037393] SCSI subsystem initialized [ 1.041366] libata version 3.00 loaded. [ 1.041403] ACPI: bus type USB registered [ 1.045616] usbcore: registered new interface driver usbfs [ 1.051373] usbcore: registered new interface driver hub [ 1.056947] usbcore: registered new device driver usb [ 1.062355] Registered efivars operations [ 1.066820] NetLabel: Initializing [ 1.070418] NetLabel: domain hash size = 128 [ 1.074997] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO [ 1.080980] NetLabel: unlabeled traffic allowed by default [ 1.086945] clocksource: Switched to clocksource arch_sys_counter [ 1.111811] VFS: Disk quotas dquot_6.6.0 [ 1.115980] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes) [ 1.123569] pnp: PnP ACPI init [ 1.127405] pnp: PnP ACPI: found 0 devices [ 1.134249] NET: Registered protocol family 2 [ 1.139069] TCP established hash table entries: 131072 (order: 4, 1048576 bytes) [ 1.147156] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes) [ 1.154513] TCP: Hash tables configured (established 131072 bind 65536) [ 1.161541] UDP hash table entries: 8192 (order: 2, 262144 bytes) [ 1.168029] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes) [ 1.175030] NET: Registered protocol family 1 [ 1.179607] PCI: CLS 64 bytes, default 128 [ 1.179708] Unpacking initramfs... [ 1.863109] Freeing initrd memory: 32640K [ 1.867699] kvm [1]: 8-bit VMID [ 1.870944] kvm [1]: IDMAP page: 40009b7000 [ 1.875301] kvm [1]: HYP VA range: 800000000000:ffffffffffff [ 1.881276] kvm [1]: Hyp mode initialized successfully [ 1.886624] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance) [ 1.895748] kvm [1]: vgic-v2@780cf000 [ 1.899646] kvm [1]: vgic interrupt IRQ1 [ 1.903757] kvm [1]: virtual timer IRQ4 [ 1.923421] audit: initializing netlink subsys (disabled) [ 1.929159] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1 [ 1.929303] Initialise system trusted keyrings [ 1.929373] workingset: timestamp_bits=44 max_order=18 bucket_order=0 [ 1.931445] zbud: loaded [ 1.931829] SELinux: Registering netfilter hooks [ 2.011199] NET: Registered protocol family 38 [ 2.015847] Key type asymmetric registered [ 2.020145] Asymmetric key parser 'x509' registered [ 2.025283] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251) [ 2.033081] io scheduler noop registered [ 2.037167] io scheduler deadline registered (default) [ 2.042536] io scheduler cfq registered [ 2.046565] io scheduler mq-deadline registered [ 2.051474] atomic64_test: passed [ 2.055189] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered. [ 2.062003] pcieport 0000:00:00.0: AER enabled with IRQ 83 [ 2.067787] pcieport 0000:00:00.0: Signaling PME with IRQ 83 [ 2.073972] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0 [ 2.081619] ACPI: Power Button [PWRB] [ 2.085922] GHES: Failed to map GSI to IRQ for generic hardware error source: 2 [ 2.093593] GHES: probe of GHES.2 failed with error 82 [ 2.098936] GHES: Failed to map GSI to IRQ for generic hardware error source: 3 [ 2.106588] GHES: probe of GHES.3 failed with error 82 [ 2.111936] GHES: Failed to map GSI to IRQ for generic hardware error source: 4 [ 2.119608] GHES: probe of GHES.4 failed with error 82 [ 2.125011] GHES: Failed to map GSI to IRQ for generic hardware error source: 5 [ 2.132659] GHES: probe of GHES.5 failed with error 82 [ 2.138017] GHES: Failed to map GSI to IRQ for generic hardware error source: 6 [ 2.145633] GHES: probe of GHES.6 failed with error 82 [ 2.151042] GHES: Failed to map GSI to IRQ for generic hardware error source: 7 [ 2.158724] GHES: probe of GHES.7 failed with error 82 [ 2.164126] GHES: Failed to map GSI to IRQ for generic hardware error source: 8 [ 2.171797] GHES: probe of GHES.8 failed with error 82 [ 2.177160] GHES: Failed to map GSI to IRQ for generic hardware error source: 9 [ 2.184845] GHES: probe of GHES.9 failed with error 82 [ 2.190233] GHES: Failed to enable APEI firmware first mode. [ 2.196297] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled [ 2.203376] console [ttyS0] disabled [ 2.227246] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A [ 2.236527] console [ttyS0] enabled [ 2.243657] bootconsole [uart8250] disabled [ 2.272506] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A [ 2.281671] cacheinfo: Unable to detect cache hierarchy for CPU 0 [ 2.287755] rdac: device handler registered [ 2.291963] hp_sw: device handler registered [ 2.296209] emc: device handler registered [ 2.300380] alua: device handler registered [ 2.304688] libphy: Fixed MDIO Bus: probed [ 2.308849] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver [ 2.315346] ehci-pci: EHCI PCI platform driver [ 2.319790] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver [ 2.325941] ohci-pci: OHCI PCI platform driver [ 2.330377] uhci_hcd: USB Universal Host Controller Interface driver [ 2.336788] xhci-hcd 808622B7:00: xHCI Host Controller [ 2.341980] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1 [ 2.349425] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010 [ 2.357756] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000 [ 2.363370] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002 [ 2.370130] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.377318] usb usb1: Product: xHCI Host Controller [ 2.382169] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd [ 2.389010] usb usb1: SerialNumber: 808622B7:00 [ 2.393727] hub 1-0:1.0: USB hub found [ 2.397479] hub 1-0:1.0: 1 port detected [ 2.401480] xhci-hcd 808622B7:00: xHCI Host Controller [ 2.406677] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2 [ 2.413986] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM. [ 2.422078] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003 [ 2.428836] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.436020] usb usb2: Product: xHCI Host Controller [ 2.440875] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd [ 2.447720] usb usb2: SerialNumber: 808622B7:00 [ 2.452398] hub 2-0:1.0: USB hub found [ 2.456137] hub 2-0:1.0: 1 port detected [ 2.460168] xhci-hcd 808622B7:01: xHCI Host Controller [ 2.465356] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3 [ 2.472759] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010 [ 2.481087] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000 [ 2.486684] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002 [ 2.493443] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.500630] usb usb3: Product: xHCI Host Controller [ 2.505482] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd [ 2.512327] usb usb3: SerialNumber: 808622B7:01 [ 2.517049] hub 3-0:1.0: USB hub found [ 2.520789] hub 3-0:1.0: 1 port detected [ 2.524790] xhci-hcd 808622B7:01: xHCI Host Controller [ 2.529976] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4 [ 2.537284] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM. [ 2.545370] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003 [ 2.552128] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.559317] usb usb4: Product: xHCI Host Controller [ 2.564168] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd [ 2.571013] usb usb4: SerialNumber: 808622B7:01 [ 2.575716] hub 4-0:1.0: USB hub found [ 2.579464] hub 4-0:1.0: 1 port detected [ 2.583505] usbcore: registered new interface driver usbserial [ 2.589324] usbcore: registered new interface driver usbserial_generic [ 2.595826] usbserial: USB Serial support registered for generic [ 2.601884] mousedev: PS/2 mouse device common for all mice [ 2.607620] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0 [ 2.613906] EFI Variables Facility v0.08 2004-May-17 [ 2.619081] hidraw: raw HID events driver (C) Jiri Kosina [ 2.624514] usbcore: registered new interface driver usbhid [ 2.630067] usbhid: USB HID core driver [ 2.633953] drop_monitor: Initializing network drop monitor service [ 2.640280] Initializing XFRM netlink socket [ 2.644691] NET: Registered protocol family 10 [ 2.649496] Segment Routing with IPv6 [ 2.653160] NET: Registered protocol family 17 [ 2.657841] registered taskstats version 1 [ 2.661921] Loading compiled-in X.509 certificates [ 2.669794] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1)) [ 2.678237] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895' [ 2.688204] zswap: loaded using pool lzo/zbud [ 2.713270] Key type big_key registered [ 2.717168] BERT: [Firmware Bug]: table invalid. [ 2.721839] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:00:41 UTC (1493395241) [ 2.757884] Freeing unused kernel memory: 1216K [ 2.767418] random: systemd: uninitialized urandom read (16 bytes read) [ 2.774589] random: systemd: uninitialized urandom read (16 bytes read) [ 2.781211] random: systemd: uninitialized urandom read (16 bytes read) [ 2.789666] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN) [ 2.807890] systemd[1]: Detected architecture arm64. [ 2.812833] systemd[1]: Running in initial RAM disk. [ 2.847031] systemd[1]: Set hostname to <localhost.localdomain>. [ 2.979499] random: systemd: uninitialized urandom read (16 bytes read) [ 2.986233] random: systemd: uninitialized urandom read (16 bytes read) [ 2.993080] random: systemd: uninitialized urandom read (16 bytes read) [ 3.000078] random: systemd: uninitialized urandom read (16 bytes read) [ 3.010094] random: systemd: uninitialized urandom read (16 bytes read) [ 3.017557] random: systemd: uninitialized urandom read (16 bytes read) [ 3.025329] random: systemd: uninitialized urandom read (16 bytes read) [ 3.087174] systemd[1]: Reached target Swap. [ 3.091495] systemd[1]: Starting Swap. [ 3.117069] systemd[1]: Reached target Timers. [ 3.121551] systemd[1]: Starting Timers. [ 3.157062] systemd[1]: Reached target Local File Systems. [ 3.162579] systemd[1]: Starting Local File Systems. [ 4.087490] device-mapper: uevent: version 1.0.3 [ 4.092424] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com [ 4.400761] libphy: APM X-Gene MDIO bus: probed [ 4.444412] pps_core: LinuxPPS API ver. 1 registered [ 4.449417] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it> [ 4.450547] xgene-ahci APMC0D0D:00: skip clock and PHY initialization [ 4.450561] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ [ 4.450629] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode [ 4.450643] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc [ 4.450670] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS [ 4.491029] sdhci: Secure Digital Host Controller Interface driver [ 4.491033] sdhci: Copyright(c) Pierre Ossman [ 4.512024] PTP clock support registered [ 4.512754] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS [ 4.525820] scsi host0: xgene-ahci [ 4.529855] scsi host1: xgene-ahci [ 4.533507] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27 [ 4.534007] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO [ 4.547673] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27 [ 4.548138] xgene-enet APMC0D05:00: clocks have been setup already [ 4.553893] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k [ 4.553894] e1000e: Copyright(c) 1999 - 2015 Intel Corporation. [ 4.554194] e1000e 0000:01:00.0: enabling device (0000 -> 0002) [ 4.554351] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode [ 4.554379] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI-X interrupts. Falling back to MSI interrupts. [ 4.554394] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI interrupts. Falling back to legacy interrupts. [ 4.617549] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock [ 4.647217] xgene-enet APMC0D30:00: clocks have been setup already [ 4.665238] xgene-ahci APMC0D0D:01: skip clock and PHY initialization [ 4.671721] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ [ 4.679178] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode [ 4.688066] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc [ 4.693143] e1000e 0000:01:00.0 eth2: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08 [ 4.693146] e1000e 0000:01:00.0 eth2: Intel(R) PRO/1000 Network Connection [ 4.693162] e1000e 0000:01:00.0 eth2: MAC: 3, PHY: 8, PBA No: E46981-008 [ 4.696649] e1000e 0000:01:00.0 enp1s0: renamed from eth2 [ 4.722971] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS [ 4.729089] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS [ 4.736197] scsi host2: xgene-ahci [ 4.740144] scsi host3: xgene-ahci [ 4.743793] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28 [ 4.751735] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28 [ 4.857119] xgene-enet APMC0D30:01: clocks have been setup already [ 4.874076] xgene-enet APMC0D31:00: clocks have been setup already [ 5.088466] ata3: SATA link down (SStatus 0 SControl 4300) [ 5.099014] ata4: SATA link down (SStatus 0 SControl 4300) [ 5.217012] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300) [ 5.223552] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133 [ 5.230179] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32) [ 5.236797] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300) [ 5.243335] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133 [ 5.249961] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32) [ 5.256600] ata1.00: configured for UDMA/133 [ 5.261153] ata2.00: configured for UDMA/133 [ 5.261162] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5 [ 5.357513] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5 [ 5.357514] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB) [ 5.357539] sd 0:0:0:0: [sda] Write Protect is off [ 5.357542] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00 [ 5.357575] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [ 5.360081] sda: sda1 sda2 sda3 [ 5.360645] sd 0:0:0:0: [sda] Attached SCSI disk [ 5.487441] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB) [ 5.494939] sd 1:0:0:0: [sdb] Write Protect is off [ 5.499723] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00 [ 5.499757] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [ 5.511144] sdb: sdb1 [ 5.513907] sd 1:0:0:0: [sdb] Attached SCSI disk [ 5.962036] random: fast init done [ 6.345919] SGI XFS with ACLs, security attributes, no debug enabled [ 6.356572] XFS (dm-0): Mounting V5 Filesystem [ 6.398617] XFS (dm-0): Ending clean mount [ 6.973554] systemd-journald[198]: Received SIGTERM from PID 1 (systemd). [ 6.989018] systemd: 20 output lines suppressed due to ratelimiting [ 7.019839] audit: type=1404 audit(1493395245.790:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295 [ 7.106438] SELinux: 32768 avtab hash slots, 105786 rules. [ 7.137061] SELinux: 32768 avtab hash slots, 105786 rules. [ 7.216426] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats [ 7.216433] SELinux: 94 classes, 105786 rules [ 7.223909] SELinux: Class sctp_socket not defined in policy. [ 7.229741] SELinux: Class icmp_socket not defined in policy. [ 7.235543] SELinux: Class ax25_socket not defined in policy. [ 7.241349] SELinux: Class ipx_socket not defined in policy. [ 7.247069] SELinux: Class netrom_socket not defined in policy. [ 7.253043] SELinux: Class atmpvc_socket not defined in policy. [ 7.259021] SELinux: Class x25_socket not defined in policy. [ 7.264736] SELinux: Class rose_socket not defined in policy. [ 7.270540] SELinux: Class decnet_socket not defined in policy. [ 7.276513] SELinux: Class atmsvc_socket not defined in policy. [ 7.282490] SELinux: Class rds_socket not defined in policy. [ 7.288207] SELinux: Class irda_socket not defined in policy. [ 7.294008] SELinux: Class pppox_socket not defined in policy. [ 7.299898] SELinux: Class llc_socket not defined in policy. [ 7.305613] SELinux: Class can_socket not defined in policy. [ 7.311330] SELinux: Class tipc_socket not defined in policy. [ 7.317135] SELinux: Class bluetooth_socket not defined in policy. [ 7.323368] SELinux: Class iucv_socket not defined in policy. [ 7.329172] SELinux: Class rxrpc_socket not defined in policy. [ 7.335059] SELinux: Class isdn_socket not defined in policy. [ 7.340863] SELinux: Class phonet_socket not defined in policy. [ 7.346836] SELinux: Class ieee802154_socket not defined in policy. [ 7.353159] SELinux: Class caif_socket not defined in policy. [ 7.358965] SELinux: Class alg_socket not defined in policy. [ 7.364679] SELinux: Class nfc_socket not defined in policy. [ 7.370397] SELinux: Class vsock_socket not defined in policy. [ 7.376284] SELinux: Class kcm_socket not defined in policy. [ 7.382002] SELinux: Class qipcrtr_socket not defined in policy. [ 7.388066] SELinux: Class smc_socket not defined in policy. [ 7.393780] SELinux: the above unknown classes and permissions will be allowed [ 7.400979] SELinux: Completing initialization. [ 7.400980] SELinux: Setting up existing superblocks. [ 7.450804] audit: type=1403 audit(1493395246.220:3): policy loaded auid=4294967295 ses=4294967295 [ 7.455100] systemd[1]: Successfully loaded SELinux policy in 435.279ms. [ 7.470770] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time. [ 7.497588] ip_tables: (C) 2000-2006 Netfilter Core Team [ 7.502902] systemd[1]: Inserted module 'ip_tables' [ 7.552579] systemd[1]: Relabelled /dev and /run in 42.787ms. [ 8.005378] systemd-journald[616]: Received request to flush runtime journal from PID 1 [ 8.167863] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered [ 8.172785] RPC: Registered named UNIX socket transport module. [ 8.172790] RPC: Registered udp transport module. [ 8.172794] RPC: Registered tcp transport module. [ 8.172797] RPC: Registered tcp NFSv4.1 backchannel transport module. [ 8.197228] xgene-rng APMC0D18:00: Couldn't get the clock for RNG [ 8.204225] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8 [ 8.213102] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered [ 8.223510] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed [ 8.289837] sd 0:0:0:0: Attached scsi generic sg0 type 0 [ 8.295454] sd 1:0:0:0: Attached scsi generic sg1 type 0 [ 8.387796] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered [ 8.518644] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS [ 8.531009] XFS (sda2): Mounting V5 Filesystem [ 8.551661] XFS (sda2): Ending clean mount [ 8.640881] Installing knfsd (copyright (C) 1996 okir@monad.swb.de). [ 8.702746] XFS (dm-2): Mounting V5 Filesystem [ 8.721803] XFS (dm-2): Ending clean mount [ 9.540607] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready [ 9.637501] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready [ 9.650581] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [ 9.656789] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [ 9.669293] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready [ 9.675510] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready [ 9.687727] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready [ 9.693923] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready [ 9.706099] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready [ 9.712318] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready [ 9.730562] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready [ 9.737469] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [ 9.744174] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready [ 9.751066] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready [ 9.757808] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready [ 10.687474] xgene-enet APMC0D30:00 eth1: Link is Down [ 10.692580] xgene-enet APMC0D05:00 eth0: Link is Down [ 10.767324] xgene-enet APMC0D30:01 eth2: Link is Down [ 12.418373] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None [ 12.426374] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready [ 13.808268] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off [ 13.815925] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready [ 108.537236] random: crng init done [ 0.000000] Booting Linux on physical CPU 0x0 [ 0.000000] Linux version 4.11.0-rc8.xgene_msix2+ (root@mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #2 SMP Fri Apr 28 12:06:14 EDT 2017 [ 0.000000] Boot CPU: AArch64 Processor [500f0001] [ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '') [ 0.000000] bootconsole [uart8250] enabled [ 0.000000] efi: Getting EFI parameters from FDT: [ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24 [ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18 [ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78. [ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000 [ 0.000000] ACPI: Early table checksum verification disabled [ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM ) [ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013) [ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724) [ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724) [ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724) [ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724) [ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724) [ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724) [ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724) [ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724) [ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724) [ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013) [ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013) [ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013) [ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013) [ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200 [ 0.000000] ACPI: NUMA: Failed to initialise from firmware [ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff] [ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0 [ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0 [ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff] [ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff] [ 0.000000] Zone ranges: [ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff] [ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff] [ 0.000000] Movable zone start for each node [ 0.000000] Early memory node ranges [ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff] [ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff] [ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff] [ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff] [ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff] [ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff] [ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff] [ 0.000000] On node 0 totalpages: 262144 [ 0.000000] DMA zone: 64 pages used for memmap [ 0.000000] DMA zone: 0 pages reserved [ 0.000000] DMA zone: 65536 pages, LIFO batch:1 [ 0.000000] Normal zone: 192 pages used for memmap [ 0.000000] Normal zone: 196608 pages, LIFO batch:1 [ 0.000000] psci: is not implemented in ACPI. [ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608 [ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536 [ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7 [ 0.000000] Detected PIPT I-cache on CPU0 [ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888 [ 0.000000] Policy zone: Normal [ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix2+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200 [ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes) [ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff] [ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved) [ 0.000000] Virtual kernel memory layout: [ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB) [ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB) [ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB) [ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB) [ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB) [ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB) [ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB) [ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB) [ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB) [ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum) [ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual) [ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB) [ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1 [ 0.000000] Hierarchical RCU implementation. [ 0.000000] Build-time adjustment of leaf fanout to 64. [ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8. [ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8 [ 0.000000] NR_IRQS:64 nr_irqs:64 0 [ 0.000000] GIC: Using split EOI/Deactivate mode [ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up [ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys). [ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns [ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns [ 0.008379] Console: colour dummy device 80x25 [ 0.012950] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000) [ 0.023693] pid_max: default: 32768 minimum: 301 [ 0.028471] ACPI: Core revision 20170119 [ 0.036063] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543) [ 0.049121] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790) [ 0.056989] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186) [ 0.063808] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543) [ 0.076952] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790) [ 0.084871] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186) [ 0.092015] ACPI: 4 ACPI AML tables successfully acquired and loaded [ 0.098567] Security Framework initialized [ 0.102838] SELinux: Initializing. [ 0.106408] SELinux: Starting in permissive mode [ 0.106607] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes) [ 0.116844] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes) [ 0.125552] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes) [ 0.132626] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes) [ 0.140570] ftrace: allocating 26815 entries in 7 pages [ 0.203748] ASID allocator initialised with 65536 entries [ 0.209911] Remapping and enabling EFI services. [ 0.214705] EFI remap 0x0000000010510000 => 0000000020000000 [ 0.220700] EFI remap 0x0000000010548000 => 0000000020018000 [ 0.226689] EFI remap 0x0000000017000000 => 0000000020020000 [ 0.232657] EFI remap 0x000000001c025000 => 0000000020035000 [ 0.238736] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware? [ 0.238741] ------------[ cut here ]------------ [ 0.250727] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104 [ 0.259532] Modules linked in: [ 0.264214] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix2+ #2 [ 0.271827] Hardware name: APM X-Gene Mustang board (DT) [ 0.277238] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000 [ 0.283324] PC is at efi_create_mapping+0x4c/0x104 [ 0.288269] LR is at efi_create_mapping+0x4c/0x104 [ 0.293127] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045 [ 0.300784] sp : ffff8003c1c03d20 [ 0.304207] x29: ffff8003c1c03d20 x28: 0000000000000000 [ 0.309655] x27: ffff000008c53000 x26: ffff000008d429c8 [ 0.315110] x25: ffff000008d42000 x24: 0000000000000000 [ 0.320556] x23: ffff000008a4c910 x22: ffff000008c60000 [ 0.325994] x21: ffff000008c53000 x20: ffff000008d429c8 [ 0.331484] x19: ffff8003f96ab348 x18: 00000000b6a2110e [ 0.337008] x17: 0000000000000000 x16: 0000000000000000 [ 0.342437] x15: ffff000008c53000 x14: 00000000cb9aed1d [ 0.347944] x13: ffff0000089ca500 x12: 0000000000040b00 [ 0.353425] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0 [ 0.358889] x9 : 0000000000000073 x8 : 6620796767756220 [ 0.364352] x7 : 2d2d20424b203436 x6 : ffff000009274718 [ 0.369825] x5 : ffff8003ffdf7530 x4 : 0000000000000000 [ 0.375280] x3 : 0000000000000000 x2 : 0000000000040b00 [ 0.380769] x1 : 0000000000040b00 x0 : 0000000000000040 [ 0.387782] ---[ end trace 9580655578fda7c1 ]--- [ 0.392519] Call trace: [ 0.395045] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60) [ 0.401735] 3b20: ffff8003f96ab348 0001000000000000 [ 0.409730] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8 [ 0.417828] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20 [ 0.425857] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8 [ 0.433946] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038 [ 0.442053] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8 [ 0.450126] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00 [ 0.458207] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530 [ 0.466227] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073 [ 0.474282] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500 [ 0.482321] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104 [ 0.488270] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238 [ 0.495238] [<ffff000008083698>] do_one_initcall+0x54/0x15c [ 0.501030] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c [ 0.507221] [<ffff0000087ae488>] kernel_init+0x18/0x108 [ 0.512606] [<ffff0000080833c0>] ret_from_fork+0x10/0x50 [ 0.518077] EFI remap 0x00000043fa7df000 => 000000002004f000 [ 0.524089] EFI remap 0x00000043fa82e000 => 000000002006e000 [ 0.530085] EFI remap 0x00000043ff9d6000 => 0000000020266000 [ 0.536148] EFI remap 0x00000043ff9ea000 => 000000002027a000 [ 0.542320] smp: Bringing up secondary CPUs ... [ 0.547164] Detected PIPT I-cache on CPU1 [ 0.547192] CPU1: Booted secondary processor [500f0001] [ 0.547411] Detected PIPT I-cache on CPU2 [ 0.547433] CPU2: Booted secondary processor [500f0001] [ 0.547646] Detected PIPT I-cache on CPU3 [ 0.547660] CPU3: Booted secondary processor [500f0001] [ 0.547854] Detected PIPT I-cache on CPU4 [ 0.547875] CPU4: Booted secondary processor [500f0001] [ 0.548076] Detected PIPT I-cache on CPU5 [ 0.548090] CPU5: Booted secondary processor [500f0001] [ 0.548281] Detected PIPT I-cache on CPU6 [ 0.548302] CPU6: Booted secondary processor [500f0001] [ 0.548505] Detected PIPT I-cache on CPU7 [ 0.548519] CPU7: Booted secondary processor [500f0001] [ 0.548562] smp: Brought up 1 node, 8 CPUs [ 0.620493] SMP: Total of 8 processors activated. [ 0.625425] CPU features: detected feature: 32-bit EL0 Support [ 0.631511] CPU: All CPU(s) started at EL2 [ 0.636531] devtmpfs: initialized [ 0.640271] SMBIOS 3.0.0 present. [ 0.643745] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016 [ 0.653110] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns [ 0.663392] futex hash table entries: 2048 (order: 2, 262144 bytes) [ 0.670383] pinctrl core: initialized pinctrl subsystem [ 0.676091] NET: Registered protocol family 16 [ 0.681082] cpuidle: using governor menu [ 0.685276] Detected 8 PCC Subspaces [ 0.689040] Registering PCC driver as Mailbox controller [ 0.694649] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000) [ 0.702382] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers. [ 0.711945] DMA: preallocated 256 KiB pool for atomic allocations [ 0.718368] ACPI: bus type PCI registered [ 0.722572] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5 [ 0.729340] Serial: AMBA PL011 UART driver [ 0.735597] HugeTLB registered 512 MB page size, pre-allocated 0 pages [ 0.743073] ACPI: Added _OSI(Module Device) [ 0.747432] ACPI: Added _OSI(Processor Device) [ 0.752083] ACPI: Added _OSI(3.0 _SCP Extensions) [ 0.757005] ACPI: Added _OSI(Processor Aggregator Device) [ 0.772090] ACPI: Interpreter enabled [ 0.775913] ACPI: Using GIC for interrupt routing [ 0.780864] ACPI: MCFG table detected, 1 entries [ 0.785858] HEST: Table parsing has been initialized. [ 0.793565] ACPI: Power Resource [SCVR] (on) [ 0.828517] xgene-msi APMC0D0E:00: APM X-Gene PCIe MSI driver loaded [ 0.835285] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff]) [ 0.841791] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI] [ 0.850585] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability] [ 0.859017] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops [ 0.871645] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace [ 0.882851] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] [ 0.891009] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window] [ 0.898264] PCI host bridge to bus 0000:00 [ 0.902574] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff]) [ 0.913134] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff]) [ 0.924779] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window] [ 0.932964] pci_bus 0000:00: root bus resource [bus 00-ff] [ 0.938697] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400 [ 0.938776] pci 0000:00:00.0: supports D1 D2 [ 0.938937] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000 [ 0.938970] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff] [ 0.938986] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff] [ 0.939003] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f] [ 0.939019] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff] [ 0.939065] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref] [ 0.939205] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold [ 0.968751] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff] [ 0.976281] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff] [ 0.982748] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff] [ 0.990222] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref] [ 0.998135] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff] [ 1.005583] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff] [ 1.013044] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f] [ 1.019447] pci 0000:00:00.0: PCI bridge to [bus 01] [ 1.024681] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff] [ 1.031099] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff] [ 1.039197] vgaarb: loaded [ 1.042181] SCSI subsystem initialized [ 1.046203] libata version 3.00 loaded. [ 1.046240] ACPI: bus type USB registered [ 1.050444] usbcore: registered new interface driver usbfs [ 1.056204] usbcore: registered new interface driver hub [ 1.061821] usbcore: registered new device driver usb [ 1.067216] Registered efivars operations [ 1.071684] NetLabel: Initializing [ 1.075256] NetLabel: domain hash size = 128 [ 1.079777] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO [ 1.085747] NetLabel: unlabeled traffic allowed by default [ 1.091744] clocksource: Switched to clocksource arch_sys_counter [ 1.116637] VFS: Disk quotas dquot_6.6.0 [ 1.120816] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes) [ 1.128374] pnp: PnP ACPI init [ 1.132213] pnp: PnP ACPI: found 0 devices [ 1.139026] NET: Registered protocol family 2 [ 1.143824] TCP established hash table entries: 131072 (order: 4, 1048576 bytes) [ 1.151892] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes) [ 1.159275] TCP: Hash tables configured (established 131072 bind 65536) [ 1.166293] UDP hash table entries: 8192 (order: 2, 262144 bytes) [ 1.172798] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes) [ 1.179867] NET: Registered protocol family 1 [ 1.184479] PCI: CLS 64 bytes, default 128 [ 1.184578] Unpacking initramfs... [ 1.868130] Freeing initrd memory: 32640K [ 1.872713] kvm [1]: 8-bit VMID [ 1.875992] kvm [1]: IDMAP page: 40009b7000 [ 1.880358] kvm [1]: HYP VA range: 800000000000:ffffffffffff [ 1.886375] kvm [1]: Hyp mode initialized successfully [ 1.891755] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance) [ 1.900915] kvm [1]: vgic-v2@780cf000 [ 1.904807] kvm [1]: vgic interrupt IRQ1 [ 1.908909] kvm [1]: virtual timer IRQ4 [ 1.928681] audit: initializing netlink subsys (disabled) [ 1.934412] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1 [ 1.934540] Initialise system trusted keyrings [ 1.934609] workingset: timestamp_bits=44 max_order=18 bucket_order=0 [ 1.936691] zbud: loaded [ 1.937062] SELinux: Registering netfilter hooks [ 2.016448] NET: Registered protocol family 38 [ 2.021088] Key type asymmetric registered [ 2.025420] Asymmetric key parser 'x509' registered [ 2.030583] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251) [ 2.038391] io scheduler noop registered [ 2.042467] io scheduler deadline registered (default) [ 2.047931] io scheduler cfq registered [ 2.051950] io scheduler mq-deadline registered [ 2.056870] atomic64_test: passed [ 2.060612] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered. [ 2.067387] pcieport 0000:00:00.0: AER enabled with IRQ 83 [ 2.073172] pcieport 0000:00:00.0: Signaling PME with IRQ 83 [ 2.079355] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0 [ 2.087026] ACPI: Power Button [PWRB] [ 2.091368] GHES: Failed to map GSI to IRQ for generic hardware error source: 2 [ 2.099018] GHES: probe of GHES.2 failed with error 82 [ 2.104386] GHES: Failed to map GSI to IRQ for generic hardware error source: 3 [ 2.112068] GHES: probe of GHES.3 failed with error 82 [ 2.117497] GHES: Failed to map GSI to IRQ for generic hardware error source: 4 [ 2.125101] GHES: probe of GHES.4 failed with error 82 [ 2.130452] GHES: Failed to map GSI to IRQ for generic hardware error source: 5 [ 2.138036] GHES: probe of GHES.5 failed with error 82 [ 2.143461] GHES: Failed to map GSI to IRQ for generic hardware error source: 6 [ 2.151121] GHES: probe of GHES.6 failed with error 82 [ 2.156510] GHES: Failed to map GSI to IRQ for generic hardware error source: 7 [ 2.164189] GHES: probe of GHES.7 failed with error 82 [ 2.169574] GHES: Failed to map GSI to IRQ for generic hardware error source: 8 [ 2.177249] GHES: probe of GHES.8 failed with error 82 [ 2.182604] GHES: Failed to map GSI to IRQ for generic hardware error source: 9 [ 2.190284] GHES: probe of GHES.9 failed with error 82 [ 2.195693] GHES: Failed to enable APEI firmware first mode. [ 2.201771] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled [ 2.208814] console [ttyS0] disabled [ 2.232712] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A [ 2.241919] console [ttyS0] enabled [ 2.249007] bootconsole [uart8250] disabled [ 2.277886] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A [ 2.287044] cacheinfo: Unable to detect cache hierarchy for CPU 0 [ 2.293126] rdac: device handler registered [ 2.297348] hp_sw: device handler registered [ 2.301595] emc: device handler registered [ 2.305766] alua: device handler registered [ 2.310077] libphy: Fixed MDIO Bus: probed [ 2.314237] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver [ 2.320734] ehci-pci: EHCI PCI platform driver [ 2.325177] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver [ 2.331327] ohci-pci: OHCI PCI platform driver [ 2.335764] uhci_hcd: USB Universal Host Controller Interface driver [ 2.342182] xhci-hcd 808622B7:00: xHCI Host Controller [ 2.347366] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1 [ 2.354792] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010 [ 2.363122] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000 [ 2.368735] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002 [ 2.375494] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.382682] usb usb1: Product: xHCI Host Controller [ 2.387533] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd [ 2.394375] usb usb1: SerialNumber: 808622B7:00 [ 2.399097] hub 1-0:1.0: USB hub found [ 2.402848] hub 1-0:1.0: 1 port detected [ 2.406851] xhci-hcd 808622B7:00: xHCI Host Controller [ 2.412034] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2 [ 2.419332] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM. [ 2.427425] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003 [ 2.434184] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.441368] usb usb2: Product: xHCI Host Controller [ 2.446223] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd [ 2.453069] usb usb2: SerialNumber: 808622B7:00 [ 2.457761] hub 2-0:1.0: USB hub found [ 2.461501] hub 2-0:1.0: 1 port detected [ 2.465534] xhci-hcd 808622B7:01: xHCI Host Controller [ 2.470715] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3 [ 2.478113] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010 [ 2.486441] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000 [ 2.492041] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002 [ 2.498793] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.505984] usb usb3: Product: xHCI Host Controller [ 2.510836] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd [ 2.517682] usb usb3: SerialNumber: 808622B7:01 [ 2.522397] hub 3-0:1.0: USB hub found [ 2.526137] hub 3-0:1.0: 1 port detected [ 2.530122] xhci-hcd 808622B7:01: xHCI Host Controller [ 2.535305] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4 [ 2.542607] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM. [ 2.550692] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003 [ 2.557451] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1 [ 2.564640] usb usb4: Product: xHCI Host Controller [ 2.569491] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd [ 2.576334] usb usb4: SerialNumber: 808622B7:01 [ 2.581037] hub 4-0:1.0: USB hub found [ 2.584788] hub 4-0:1.0: 1 port detected [ 2.588828] usbcore: registered new interface driver usbserial [ 2.594649] usbcore: registered new interface driver usbserial_generic [ 2.601150] usbserial: USB Serial support registered for generic [ 2.607208] mousedev: PS/2 mouse device common for all mice [ 2.612948] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0 [ 2.619241] EFI Variables Facility v0.08 2004-May-17 [ 2.624405] hidraw: raw HID events driver (C) Jiri Kosina [ 2.629836] usbcore: registered new interface driver usbhid [ 2.635388] usbhid: USB HID core driver [ 2.639274] drop_monitor: Initializing network drop monitor service [ 2.645602] Initializing XFRM netlink socket [ 2.650006] NET: Registered protocol family 10 [ 2.654800] Segment Routing with IPv6 [ 2.658462] NET: Registered protocol family 17 [ 2.663153] registered taskstats version 1 [ 2.667231] Loading compiled-in X.509 certificates [ 2.675102] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1)) [ 2.683548] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895' [ 2.693521] zswap: loaded using pool lzo/zbud [ 2.717703] Key type big_key registered [ 2.721594] BERT: [Firmware Bug]: table invalid. [ 2.726272] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:24:34 UTC (1493396674) [ 2.759039] Freeing unused kernel memory: 1216K [ 2.768550] random: systemd: uninitialized urandom read (16 bytes read) [ 2.775743] random: systemd: uninitialized urandom read (16 bytes read) [ 2.782366] random: systemd: uninitialized urandom read (16 bytes read) [ 2.790811] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN) [ 2.809044] systemd[1]: Detected architecture arm64. [ 2.814004] systemd[1]: Running in initial RAM disk. [ 2.851832] systemd[1]: Set hostname to <localhost.localdomain>. [ 2.970013] random: systemd: uninitialized urandom read (16 bytes read) [ 2.976652] random: systemd: uninitialized urandom read (16 bytes read) [ 2.983295] random: systemd: uninitialized urandom read (16 bytes read) [ 2.989979] random: systemd: uninitialized urandom read (16 bytes read) [ 2.997677] random: systemd: uninitialized urandom read (16 bytes read) [ 3.004524] random: systemd: uninitialized urandom read (16 bytes read) [ 3.011427] random: systemd: uninitialized urandom read (16 bytes read) [ 3.051810] systemd[1]: Reached target Timers. [ 3.056250] systemd[1]: Starting Timers. [ 3.091786] systemd[1]: Created slice Root Slice. [ 3.096483] systemd[1]: Starting Root Slice. [ 3.131780] systemd[1]: Listening on udev Kernel Socket. [ 3.137083] systemd[1]: Starting udev Kernel Socket. [ 4.102501] device-mapper: uevent: version 1.0.3 [ 4.107414] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com [ 4.412199] xgene-ahci APMC0D0D:00: skip clock and PHY initialization [ 4.418619] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ [ 4.426125] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode [ 4.434987] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc [ 4.434998] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS [ 4.470653] pps_core: LinuxPPS API ver. 1 registered [ 4.475669] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it> [ 4.491986] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS [ 4.507826] PTP clock support registered [ 4.521307] sdhci: Secure Digital Host Controller Interface driver [ 4.527512] sdhci: Copyright(c) Pierre Ossman [ 4.577139] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO [ 4.612188] scsi host0: xgene-ahci [ 4.615838] scsi host1: xgene-ahci [ 4.619314] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27 [ 4.621598] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k [ 4.621602] e1000e: Copyright(c) 1999 - 2015 Intel Corporation. [ 4.621841] e1000e 0000:01:00.0: enabling device (0000 -> 0002) [ 4.622056] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode [ 4.654091] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27 [ 4.662326] xgene-ahci APMC0D0D:01: skip clock and PHY initialization [ 4.668755] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ [ 4.676203] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode [ 4.685069] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc [ 4.693201] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS [ 4.693880] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock [ 4.706842] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS [ 4.714376] scsi host2: xgene-ahci [ 4.718239] scsi host3: xgene-ahci [ 4.721961] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28 [ 4.729850] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28 [ 4.738769] libphy: APM X-Gene MDIO bus: probed [ 4.791711] e1000e 0000:01:00.0 eth0: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08 [ 4.799622] e1000e 0000:01:00.0 eth0: Intel(R) PRO/1000 Network Connection [ 4.806511] e1000e 0000:01:00.0 eth0: MAC: 3, PHY: 8, PBA No: E46981-008 [ 4.808607] xgene-enet APMC0D05:00: clocks have been setup already [ 4.821883] e1000e 0000:01:00.0 enp1s0: renamed from eth0 [ 5.011899] xgene-enet APMC0D30:00: clocks have been setup already [ 5.063242] ata3: SATA link down (SStatus 0 SControl 4300) [ 5.068752] ata4: SATA link down (SStatus 0 SControl 4300) [ 5.121895] xgene-enet APMC0D30:01: clocks have been setup already [ 5.136261] xgene-enet APMC0D31:00: clocks have been setup already [ 5.331836] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300) [ 5.338118] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300) [ 5.344486] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133 [ 5.351098] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32) [ 5.357813] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133 [ 5.364432] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32) [ 5.371105] ata2.00: configured for UDMA/133 [ 5.375470] ata1.00: configured for UDMA/133 [ 5.380021] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5 [ 5.482145] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB) [ 5.482607] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5 [ 5.498165] sd 0:0:0:0: [sda] Write Protect is off [ 5.502945] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00 [ 5.502986] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [ 5.514451] sda: sda1 sda2 sda3 [ 5.518153] sd 0:0:0:0: [sda] Attached SCSI disk [ 5.602396] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB) [ 5.609881] sd 1:0:0:0: [sdb] Write Protect is off [ 5.614681] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00 [ 5.614739] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [ 5.626189] sdb: sdb1 [ 5.628980] sd 1:0:0:0: [sdb] Attached SCSI disk [ 6.113010] random: fast init done [ 6.461233] SGI XFS with ACLs, security attributes, no debug enabled [ 6.471638] XFS (dm-0): Mounting V5 Filesystem [ 6.501907] XFS (dm-0): Ending clean mount [ 7.216272] systemd-journald[201]: Received SIGTERM from PID 1 (systemd). [ 7.263513] systemd: 20 output lines suppressed due to ratelimiting [ 7.301701] audit: type=1404 audit(1493396679.060:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295 [ 7.410911] SELinux: 32768 avtab hash slots, 105786 rules. [ 7.516206] SELinux: 32768 avtab hash slots, 105786 rules. [ 7.718524] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats [ 7.718542] SELinux: 94 classes, 105786 rules [ 7.741338] SELinux: Class sctp_socket not defined in policy. [ 7.747177] SELinux: Class icmp_socket not defined in policy. [ 7.753000] SELinux: Class ax25_socket not defined in policy. [ 7.758809] SELinux: Class ipx_socket not defined in policy. [ 7.764542] SELinux: Class netrom_socket not defined in policy. [ 7.770524] SELinux: Class atmpvc_socket not defined in policy. [ 7.776515] SELinux: Class x25_socket not defined in policy. [ 7.782247] SELinux: Class rose_socket not defined in policy. [ 7.788056] SELinux: Class decnet_socket not defined in policy. [ 7.794048] SELinux: Class atmsvc_socket not defined in policy. [ 7.800028] SELinux: Class rds_socket not defined in policy. [ 7.805760] SELinux: Class irda_socket not defined in policy. [ 7.811568] SELinux: Class pppox_socket not defined in policy. [ 7.817473] SELinux: Class llc_socket not defined in policy. [ 7.823204] SELinux: Class can_socket not defined in policy. [ 7.828926] SELinux: Class tipc_socket not defined in policy. [ 7.834744] SELinux: Class bluetooth_socket not defined in policy. [ 7.840984] SELinux: Class iucv_socket not defined in policy. [ 7.846802] SELinux: Class rxrpc_socket not defined in policy. [ 7.852707] SELinux: Class isdn_socket not defined in policy. [ 7.858514] SELinux: Class phonet_socket not defined in policy. [ 7.864504] SELinux: Class ieee802154_socket not defined in policy. [ 7.870831] SELinux: Class caif_socket not defined in policy. [ 7.876649] SELinux: Class alg_socket not defined in policy. [ 7.882381] SELinux: Class nfc_socket not defined in policy. [ 7.888103] SELinux: Class vsock_socket not defined in policy. [ 7.894008] SELinux: Class kcm_socket not defined in policy. [ 7.899730] SELinux: Class qipcrtr_socket not defined in policy. [ 7.905822] SELinux: Class smc_socket not defined in policy. [ 7.911545] SELinux: the above unknown classes and permissions will be allowed [ 7.918772] SELinux: Completing initialization. [ 7.918776] SELinux: Setting up existing superblocks. [ 7.993509] audit: type=1403 audit(1493396679.760:3): policy loaded auid=4294967295 ses=4294967295 [ 8.002331] systemd[1]: Successfully loaded SELinux policy in 701.052ms. [ 8.029233] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time. [ 8.076262] ip_tables: (C) 2000-2006 Netfilter Core Team [ 8.081619] systemd[1]: Inserted module 'ip_tables' [ 8.199798] systemd[1]: Relabelled /dev and /run in 108.385ms. [ 8.666128] systemd-journald[618]: Received request to flush runtime journal from PID 1 [ 8.793125] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed [ 8.809689] RPC: Registered named UNIX socket transport module. [ 8.815634] RPC: Registered udp transport module. [ 8.820316] RPC: Registered tcp transport module. [ 8.825018] RPC: Registered tcp NFSv4.1 backchannel transport module. [ 8.872038] sd 0:0:0:0: Attached scsi generic sg0 type 0 [ 8.877502] sd 1:0:0:0: Attached scsi generic sg1 type 0 [ 8.883774] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered [ 8.892777] xgene-rng APMC0D18:00: Couldn't get the clock for RNG [ 8.899408] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered [ 8.910481] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8 [ 8.918470] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered [ 9.044885] Installing knfsd (copyright (C) 1996 okir@monad.swb.de). [ 9.118751] XFS (sda2): Mounting V5 Filesystem [ 9.125864] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS [ 9.141179] XFS (sda2): Ending clean mount [ 9.286221] XFS (dm-2): Mounting V5 Filesystem [ 9.305151] XFS (dm-2): Ending clean mount [ 10.242361] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready [ 10.339596] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready [ 10.367672] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [ 10.374758] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [ 10.403220] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready [ 10.410389] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready [ 10.437608] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready [ 10.444895] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready [ 10.471938] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready [ 10.478809] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready [ 10.562700] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready [ 10.571438] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [ 10.580052] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready [ 10.588624] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready [ 10.597191] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready [ 11.412166] xgene-enet APMC0D05:00 eth0: Link is Down [ 11.492472] xgene-enet APMC0D30:00 eth1: Link is Down [ 11.497600] xgene-enet APMC0D30:01 eth2: Link is Down [ 14.533303] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None [ 14.541337] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready [ 14.548556] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off [ 14.556244] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
On 04/28/2017 12:36 PM, Jon Masters wrote: > On 04/28/2017 01:11 AM, Jon Masters wrote: >> On 04/27/2017 08:54 PM, Khuong Dinh wrote: >>> From: Khuong Dinh <kdinh@apm.com> >>> >>> This patch makes pci-xgene-msi driver ACPI-aware and provides >>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >>> >>> Signed-off-by: Khuong Dinh <kdinh@apm.com> >>> Signed-off-by: Duc Dang <dhdang@apm.com> >>> Acked-by: Marc Zyngier <marc.zyngier@arm.com> >> >> Thanks. Currently testing. Will review and followup shortly. > > Attaching boot logs before and after. I tested with an Intel e1000. > Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI > interrupts working. Please followup with answers to Marc's question > about how you propose to guarantee boottime ordering of binding. Any word? Jon.
Hi Marc, There's no explicit dependency between pcie driver and msi controller. The current solution that we did is relying on the node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a higher priority in start ordering. This method could be applied in case of msi and pcie are the same level of subsys_init (in ACPI boot). However, PCIe driver has not supported for this dependency check yet. How do you think about this solution. Best regards, Khuong On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: > On 28/04/17 01:54, Khuong Dinh wrote: >> From: Khuong Dinh <kdinh@apm.com> >> >> This patch makes pci-xgene-msi driver ACPI-aware and provides >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> >> Signed-off-by: Duc Dang <dhdang@apm.com> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> >> --- >> v2: >> - Verify with BIOS version 3.06.25 and 3.07.09 >> v1: >> - Initial version >> --- >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- >> 1 files changed, 32 insertions(+), 3 deletions(-) >> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c >> index f1b633b..00aaa3d 100644 >> --- a/drivers/pci/host/pci-xgene-msi.c >> +++ b/drivers/pci/host/pci-xgene-msi.c >> @@ -24,6 +24,7 @@ >> #include <linux/pci.h> >> #include <linux/platform_device.h> >> #include <linux/of_pci.h> >> +#include <linux/acpi.h> >> >> #define MSI_IR0 0x000000 >> #define MSI_INT0 0x800000 >> @@ -39,7 +40,7 @@ struct xgene_msi_group { >> }; >> >> struct xgene_msi { >> - struct device_node *node; >> + struct fwnode_handle *fwnode; >> struct irq_domain *inner_domain; >> struct irq_domain *msi_domain; >> u64 msi_addr; >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, >> .free = xgene_irq_domain_free, >> }; >> >> +#ifdef CONFIG_ACPI >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) >> +{ >> + return xgene_msi_ctrl.fwnode; >> +} >> +#endif >> + >> static int xgene_allocate_domains(struct xgene_msi *msi) >> { >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> if (!msi->inner_domain) >> return -ENOMEM; >> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, >> &xgene_msi_domain_info, >> msi->inner_domain); >> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> return -ENOMEM; >> } >> >> +#ifdef CONFIG_ACPI >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); >> +#endif >> return 0; >> } >> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) >> {}, >> }; >> >> +#ifdef CONFIG_ACPI >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { >> + {"APMC0D0E", 0}, >> + { }, >> +}; >> +#endif >> + >> static int xgene_msi_probe(struct platform_device *pdev) >> { >> struct resource *res; >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) >> goto error; >> } >> xgene_msi->msi_addr = res->start; >> - xgene_msi->node = pdev->dev.of_node; >> + >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); >> + if (!xgene_msi->fwnode) { >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); > > Please provide something other than NULL, such as the base address if > the device. That's quite useful for debugging. > >> + if (!xgene_msi->fwnode) { >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); >> + rc = ENOMEM; >> + goto error; >> + } >> + } >> + >> xgene_msi->num_cpus = num_possible_cpus(); >> >> rc = xgene_msi_init_allocator(xgene_msi); >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) >> .driver = { >> .name = "xgene-msi", >> .of_match_table = xgene_msi_match_table, >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), >> }, >> .probe = xgene_msi_probe, >> .remove = xgene_msi_remove, >> > > The code is trivial, but relies on the MSI controller to probe before > the PCI bus. What enforces this? How is it making sure that this is not > going to break in the next kernel release? As far as I can tell, there > is no explicit dependency between the two, making it the whole thing > extremely fragile. > > Thanks, > > M. > -- > Jazz is not dead. It just smells funny...
On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: > Hi Marc, > There's no explicit dependency between pcie driver and msi > controller. The current solution that we did is relying on the > node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a > higher priority in start ordering. This method could be applied in > case of msi and pcie are the same level of subsys_init (in ACPI > boot). However, PCIe driver has not supported for this dependency > check yet. How do you think about this solution. First off, you can't post emails with confidentiality notices on mailing lists so remove it from now onwards. Secondly, I commented on this before, so you know what my opinion is. Finally, please execute this command on the vmlinux that "works" for you: nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' Even by ordering devices in the ACPI tables (that I abhor) I still do not understand how this works (I mean without relying on kernel link order to ensure that the MSI driver is probed before PCI devices are enumerated in acpi_init()). Thanks, Lorenzo > Best regards, > Khuong > > On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: > > On 28/04/17 01:54, Khuong Dinh wrote: > >> From: Khuong Dinh <kdinh@apm.com> > >> > >> This patch makes pci-xgene-msi driver ACPI-aware and provides > >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > >> > >> Signed-off-by: Khuong Dinh <kdinh@apm.com> > >> Signed-off-by: Duc Dang <dhdang@apm.com> > >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> > >> --- > >> v2: > >> - Verify with BIOS version 3.06.25 and 3.07.09 > >> v1: > >> - Initial version > >> --- > >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- > >> 1 files changed, 32 insertions(+), 3 deletions(-) > >> > >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c > >> index f1b633b..00aaa3d 100644 > >> --- a/drivers/pci/host/pci-xgene-msi.c > >> +++ b/drivers/pci/host/pci-xgene-msi.c > >> @@ -24,6 +24,7 @@ > >> #include <linux/pci.h> > >> #include <linux/platform_device.h> > >> #include <linux/of_pci.h> > >> +#include <linux/acpi.h> > >> > >> #define MSI_IR0 0x000000 > >> #define MSI_INT0 0x800000 > >> @@ -39,7 +40,7 @@ struct xgene_msi_group { > >> }; > >> > >> struct xgene_msi { > >> - struct device_node *node; > >> + struct fwnode_handle *fwnode; > >> struct irq_domain *inner_domain; > >> struct irq_domain *msi_domain; > >> u64 msi_addr; > >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, > >> .free = xgene_irq_domain_free, > >> }; > >> > >> +#ifdef CONFIG_ACPI > >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) > >> +{ > >> + return xgene_msi_ctrl.fwnode; > >> +} > >> +#endif > >> + > >> static int xgene_allocate_domains(struct xgene_msi *msi) > >> { > >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, > >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >> if (!msi->inner_domain) > >> return -ENOMEM; > >> > >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), > >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, > >> &xgene_msi_domain_info, > >> msi->inner_domain); > >> > >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >> return -ENOMEM; > >> } > >> > >> +#ifdef CONFIG_ACPI > >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); > >> +#endif > >> return 0; > >> } > >> > >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) > >> {}, > >> }; > >> > >> +#ifdef CONFIG_ACPI > >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { > >> + {"APMC0D0E", 0}, > >> + { }, > >> +}; > >> +#endif > >> + > >> static int xgene_msi_probe(struct platform_device *pdev) > >> { > >> struct resource *res; > >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) > >> goto error; > >> } > >> xgene_msi->msi_addr = res->start; > >> - xgene_msi->node = pdev->dev.of_node; > >> + > >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); > >> + if (!xgene_msi->fwnode) { > >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); > > > > Please provide something other than NULL, such as the base address if > > the device. That's quite useful for debugging. > > > >> + if (!xgene_msi->fwnode) { > >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); > >> + rc = ENOMEM; > >> + goto error; > >> + } > >> + } > >> + > >> xgene_msi->num_cpus = num_possible_cpus(); > >> > >> rc = xgene_msi_init_allocator(xgene_msi); > >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) > >> .driver = { > >> .name = "xgene-msi", > >> .of_match_table = xgene_msi_match_table, > >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), > >> }, > >> .probe = xgene_msi_probe, > >> .remove = xgene_msi_remove, > >> > > > > The code is trivial, but relies on the MSI controller to probe before > > the PCI bus. What enforces this? How is it making sure that this is not > > going to break in the next kernel release? As far as I can tell, there > > is no explicit dependency between the two, making it the whole thing > > extremely fragile. > > > > Thanks, > > > > M. > > -- > > Jazz is not dead. It just smells funny... > > -- > CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is > for the sole use of the intended recipient(s) and contains information that > is confidential and proprietary to Applied Micro Circuits Corporation or > its subsidiaries. It is to be used solely for the purpose of furthering the > parties' business relationship. All unauthorized review, use, disclosure or > distribution is prohibited. If you are not the intended recipient, please > contact the sender by reply e-mail and destroy all copies of the original > message.
Hi Lorenzo, On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote: > On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: >> Hi Marc, >> There's no explicit dependency between pcie driver and msi >> controller. The current solution that we did is relying on the >> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a >> higher priority in start ordering. This method could be applied in >> case of msi and pcie are the same level of subsys_init (in ACPI >> boot). However, PCIe driver has not supported for this dependency >> check yet. How do you think about this solution. > > First off, you can't post emails with confidentiality notices on > mailing lists so remove it from now onwards. Fixed > Secondly, I commented on this before, so you know what my opinion is. I got your opinion. I'll implement as your suggestion. > Finally, please execute this command on the vmlinux that "works" > for you: > > nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' ffff000008dab2d8 t __initcall_acpi_init4 ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 Best regards, Khuong Dinh > Even by ordering devices in the ACPI tables (that I abhor) I still do > not understand how this works (I mean without relying on kernel link > order to ensure that the MSI driver is probed before PCI devices are > enumerated in acpi_init()). > > Thanks, > Lorenzo > >> Best regards, >> Khuong >> >> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: >> > On 28/04/17 01:54, Khuong Dinh wrote: >> >> From: Khuong Dinh <kdinh@apm.com> >> >> >> >> This patch makes pci-xgene-msi driver ACPI-aware and provides >> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >> >> >> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> >> >> Signed-off-by: Duc Dang <dhdang@apm.com> >> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> >> >> --- >> >> v2: >> >> - Verify with BIOS version 3.06.25 and 3.07.09 >> >> v1: >> >> - Initial version >> >> --- >> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- >> >> 1 files changed, 32 insertions(+), 3 deletions(-) >> >> >> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c >> >> index f1b633b..00aaa3d 100644 >> >> --- a/drivers/pci/host/pci-xgene-msi.c >> >> +++ b/drivers/pci/host/pci-xgene-msi.c >> >> @@ -24,6 +24,7 @@ >> >> #include <linux/pci.h> >> >> #include <linux/platform_device.h> >> >> #include <linux/of_pci.h> >> >> +#include <linux/acpi.h> >> >> >> >> #define MSI_IR0 0x000000 >> >> #define MSI_INT0 0x800000 >> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { >> >> }; >> >> >> >> struct xgene_msi { >> >> - struct device_node *node; >> >> + struct fwnode_handle *fwnode; >> >> struct irq_domain *inner_domain; >> >> struct irq_domain *msi_domain; >> >> u64 msi_addr; >> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, >> >> .free = xgene_irq_domain_free, >> >> }; >> >> >> >> +#ifdef CONFIG_ACPI >> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) >> >> +{ >> >> + return xgene_msi_ctrl.fwnode; >> >> +} >> >> +#endif >> >> + >> >> static int xgene_allocate_domains(struct xgene_msi *msi) >> >> { >> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, >> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> >> if (!msi->inner_domain) >> >> return -ENOMEM; >> >> >> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), >> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, >> >> &xgene_msi_domain_info, >> >> msi->inner_domain); >> >> >> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> >> return -ENOMEM; >> >> } >> >> >> >> +#ifdef CONFIG_ACPI >> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); >> >> +#endif >> >> return 0; >> >> } >> >> >> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) >> >> {}, >> >> }; >> >> >> >> +#ifdef CONFIG_ACPI >> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { >> >> + {"APMC0D0E", 0}, >> >> + { }, >> >> +}; >> >> +#endif >> >> + >> >> static int xgene_msi_probe(struct platform_device *pdev) >> >> { >> >> struct resource *res; >> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) >> >> goto error; >> >> } >> >> xgene_msi->msi_addr = res->start; >> >> - xgene_msi->node = pdev->dev.of_node; >> >> + >> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); >> >> + if (!xgene_msi->fwnode) { >> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); >> > >> > Please provide something other than NULL, such as the base address if >> > the device. That's quite useful for debugging. >> > >> >> + if (!xgene_msi->fwnode) { >> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); >> >> + rc = ENOMEM; >> >> + goto error; >> >> + } >> >> + } >> >> + >> >> xgene_msi->num_cpus = num_possible_cpus(); >> >> >> >> rc = xgene_msi_init_allocator(xgene_msi); >> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) >> >> .driver = { >> >> .name = "xgene-msi", >> >> .of_match_table = xgene_msi_match_table, >> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), >> >> }, >> >> .probe = xgene_msi_probe, >> >> .remove = xgene_msi_remove, >> >> >> > >> > The code is trivial, but relies on the MSI controller to probe before >> > the PCI bus. What enforces this? How is it making sure that this is not >> > going to break in the next kernel release? As far as I can tell, there >> > is no explicit dependency between the two, making it the whole thing >> > extremely fragile. >> > >> > Thanks, >> > >> > M. >> > -- >> > Jazz is not dead. It just smells funny... >> >> -- >> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is >> for the sole use of the intended recipient(s) and contains information that >> is confidential and proprietary to Applied Micro Circuits Corporation or >> its subsidiaries. It is to be used solely for the purpose of furthering the >> parties' business relationship. All unauthorized review, use, disclosure or >> distribution is prohibited. If you are not the intended recipient, please >> contact the sender by reply e-mail and destroy all copies of the original >> message.
On 04/27/2017 08:54 PM, Khuong Dinh wrote: > From: Khuong Dinh <kdinh@apm.com> > > This patch makes pci-xgene-msi driver ACPI-aware and provides > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > > Signed-off-by: Khuong Dinh <kdinh@apm.com> > Signed-off-by: Duc Dang <dhdang@apm.com> > Acked-by: Marc Zyngier <marc.zyngier@arm.com> > --- > v2: > - Verify with BIOS version 3.06.25 and 3.07.09 > v1: > - Initial version Hoping I just missed it, but was a follow-up (v3) ever provided for this? It seems this thread ended with the expectation that a v3 would be provided to address the sequencing issue. Regards, -Jon
Hi Jon, > > From: Khuong Dinh <kdinh@apm.com> > > > > This patch makes pci-xgene-msi driver ACPI-aware and provides > > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > > > > Signed-off-by: Khuong Dinh <kdinh@apm.com> > > Signed-off-by: Duc Dang <dhdang@apm.com> > > Acked-by: Marc Zyngier <marc.zyngier@arm.com> > > --- > > v2: > > - Verify with BIOS version 3.06.25 and 3.07.09 > > v1: > > - Initial version > > > Hoping I just missed it, but was a follow-up (v3) ever provided for > this? It seems this thread ended with the expectation that a v3 would be > provided to address the sequencing issue. No... You didn't missed anything. Khuong is out last couple days. He is currently implementing the dependence requirement using the _DED method. As soon as he get this working correctly, he will post the next version. Though, there isn't anything wrong with the last version besides future potential change in the kernel which is very unlikely to change for the next one or more release of the future kernel. -Loc
Hi Lorenzo, On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote: > Hi Lorenzo, > > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi > <lorenzo.pieralisi@arm.com> wrote: >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: >>> Hi Marc, >>> There's no explicit dependency between pcie driver and msi >>> controller. The current solution that we did is relying on the >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a >>> higher priority in start ordering. This method could be applied in >>> case of msi and pcie are the same level of subsys_init (in ACPI >>> boot). However, PCIe driver has not supported for this dependency >>> check yet. How do you think about this solution. >> >> First off, you can't post emails with confidentiality notices on >> mailing lists so remove it from now onwards. > > Fixed > >> Secondly, I commented on this before, so you know what my opinion is. > > I got your opinion. I'll implement as your suggestion. > Regarding to your previous suggestion to add a hook walking the ACPI namespace before acpi_bus_scan() in acpi_scan_init() to make MSI controllers must be probed before PCI. I have a concern that the MSI namespace “ _SB.MSIX” is not a unique name and how can we walk the ACPI namespace and use this name to make MSI probed before PCI. May you have little bit more information for this or do you have another suggestion for this? There’s another solution which makes this simpler and I’d like to ask your opinion about this. The solution is to make an hierarchy between MSI and PCI nodes as below: Device(\_SB.MSIX) { Device(\_SB.PCI0) Device(\_SB.PCI1) …… } In other word, MSIX node is a parent node of PCI nodes in ACPI table. In this sense, there’s an explicit dependency between MSI and PCI, MSI controller must be probed before PCI and it will guarantee not breaking next kernel releases. How do you think about this solution. I also tried to use _DEP method to describe the dependency of PCIe nodes, but it looks that it does not work as PCI and MSI are handed by acpi_bus_scan and still having issue when we re-probe PCI. Thanks, Khuong Dinh >> Finally, please execute this command on the vmlinux that "works" >> for you: >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > ffff000008dab2d8 t __initcall_acpi_init4 > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 > > Best regards, > Khuong Dinh > >> Even by ordering devices in the ACPI tables (that I abhor) I still do >> not understand how this works (I mean without relying on kernel link >> order to ensure that the MSI driver is probed before PCI devices are >> enumerated in acpi_init()). >> >> Thanks, >> Lorenzo >> >>> Best regards, >>> Khuong >>> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: >>> > On 28/04/17 01:54, Khuong Dinh wrote: >>> >> From: Khuong Dinh <kdinh@apm.com> >>> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >>> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> >>> >> --- >>> >> v2: >>> >> - Verify with BIOS version 3.06.25 and 3.07.09 >>> >> v1: >>> >> - Initial version >>> >> --- >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- >>> >> 1 files changed, 32 insertions(+), 3 deletions(-) >>> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c >>> >> index f1b633b..00aaa3d 100644 >>> >> --- a/drivers/pci/host/pci-xgene-msi.c >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c >>> >> @@ -24,6 +24,7 @@ >>> >> #include <linux/pci.h> >>> >> #include <linux/platform_device.h> >>> >> #include <linux/of_pci.h> >>> >> +#include <linux/acpi.h> >>> >> >>> >> #define MSI_IR0 0x000000 >>> >> #define MSI_INT0 0x800000 >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { >>> >> }; >>> >> >>> >> struct xgene_msi { >>> >> - struct device_node *node; >>> >> + struct fwnode_handle *fwnode; >>> >> struct irq_domain *inner_domain; >>> >> struct irq_domain *msi_domain; >>> >> u64 msi_addr; >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, >>> >> .free = xgene_irq_domain_free, >>> >> }; >>> >> >>> >> +#ifdef CONFIG_ACPI >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) >>> >> +{ >>> >> + return xgene_msi_ctrl.fwnode; >>> >> +} >>> >> +#endif >>> >> + >>> >> static int xgene_allocate_domains(struct xgene_msi *msi) >>> >> { >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >>> >> if (!msi->inner_domain) >>> >> return -ENOMEM; >>> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, >>> >> &xgene_msi_domain_info, >>> >> msi->inner_domain); >>> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >>> >> return -ENOMEM; >>> >> } >>> >> >>> >> +#ifdef CONFIG_ACPI >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); >>> >> +#endif >>> >> return 0; >>> >> } >>> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) >>> >> {}, >>> >> }; >>> >> >>> >> +#ifdef CONFIG_ACPI >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { >>> >> + {"APMC0D0E", 0}, >>> >> + { }, >>> >> +}; >>> >> +#endif >>> >> + >>> >> static int xgene_msi_probe(struct platform_device *pdev) >>> >> { >>> >> struct resource *res; >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) >>> >> goto error; >>> >> } >>> >> xgene_msi->msi_addr = res->start; >>> >> - xgene_msi->node = pdev->dev.of_node; >>> >> + >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); >>> >> + if (!xgene_msi->fwnode) { >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); >>> > >>> > Please provide something other than NULL, such as the base address if >>> > the device. That's quite useful for debugging. >>> > >>> >> + if (!xgene_msi->fwnode) { >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); >>> >> + rc = ENOMEM; >>> >> + goto error; >>> >> + } >>> >> + } >>> >> + >>> >> xgene_msi->num_cpus = num_possible_cpus(); >>> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi); >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) >>> >> .driver = { >>> >> .name = "xgene-msi", >>> >> .of_match_table = xgene_msi_match_table, >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), >>> >> }, >>> >> .probe = xgene_msi_probe, >>> >> .remove = xgene_msi_remove, >>> >> >>> > >>> > The code is trivial, but relies on the MSI controller to probe before >>> > the PCI bus. What enforces this? How is it making sure that this is not >>> > going to break in the next kernel release? As far as I can tell, there >>> > is no explicit dependency between the two, making it the whole thing >>> > extremely fragile. >>> > >>> > Thanks, >>> > >>> > M. >>> > -- >>> > Jazz is not dead. It just smells funny... >>> >>> --
On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote: > Hi Lorenzo, > > On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote: > > Hi Lorenzo, > > > > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi > > <lorenzo.pieralisi@arm.com> wrote: > >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: > >>> Hi Marc, > >>> There's no explicit dependency between pcie driver and msi > >>> controller. The current solution that we did is relying on the > >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a > >>> higher priority in start ordering. This method could be applied in > >>> case of msi and pcie are the same level of subsys_init (in ACPI > >>> boot). However, PCIe driver has not supported for this dependency > >>> check yet. How do you think about this solution. > >> > >> First off, you can't post emails with confidentiality notices on > >> mailing lists so remove it from now onwards. > > > > Fixed > > > >> Secondly, I commented on this before, so you know what my opinion is. > > > > I got your opinion. I'll implement as your suggestion. > > > > Regarding to your previous suggestion to add a hook walking the ACPI > namespace before acpi_bus_scan() in acpi_scan_init() to make MSI > controllers must be probed before PCI. I have a concern that the > MSI namespace “ _SB.MSIX” is not a unique name and how can we walk > the ACPI namespace and use this name to make MSI probed before PCI. > May you have little bit more information for this or do you have > another suggestion for this? > > There’s another solution which makes this simpler and I’d like to > ask your opinion about this. > The solution is to make an hierarchy between MSI and PCI nodes as below: > Device(\_SB.MSIX) { > Device(\_SB.PCI0) > Device(\_SB.PCI1) > …… > } > In other word, MSIX node is a parent node of PCI nodes in ACPI > table. In this sense, there’s an explicit dependency between MSI > and PCI, MSI controller must be probed before PCI and it will > guarantee not breaking next kernel releases. How do you think about > this solution. I think that's a plaster as ineffective as reordering nodes, in short it is not a solution and you still rely on kernel link ordering, you can fiddle with ACPI tables as much as you want but that does not change. > I also tried to use _DEP method to describe the dependency of PCIe > nodes, but it looks that it does not work as PCI and MSI are handed > by acpi_bus_scan and still having issue when we re-probe PCI. That's a tad vague to be frank, anyway it is pretty clear to me that we have hit a wall. In ACPI there is no way to describe probe dependencies like the one you have, it is as simple as that, and this MSI issue you are facing is just an example, there are more eg: https://www.spinics.net/lists/arm-kernel/msg585825.html At the end of the day the choice is simple either we accept (and we maintain because that's the problem) these hacks - and I am not willing to do that - or we find a way to solve this from a general perspective not as a point hack. I can have a look at solving the whole issue but it won't happen tomorrow. Thanks, Lorenzo > Thanks, > Khuong Dinh > > >> Finally, please execute this command on the vmlinux that "works" > >> for you: > >> > >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > > > > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > > ffff000008dab2d8 t __initcall_acpi_init4 > > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 > > > > Best regards, > > Khuong Dinh > > > >> Even by ordering devices in the ACPI tables (that I abhor) I still do > >> not understand how this works (I mean without relying on kernel link > >> order to ensure that the MSI driver is probed before PCI devices are > >> enumerated in acpi_init()). > >> > >> Thanks, > >> Lorenzo > >> > >>> Best regards, > >>> Khuong > >>> > >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: > >>> > On 28/04/17 01:54, Khuong Dinh wrote: > >>> >> From: Khuong Dinh <kdinh@apm.com> > >>> >> > >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides > >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > >>> >> > >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> > >>> >> Signed-off-by: Duc Dang <dhdang@apm.com> > >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> > >>> >> --- > >>> >> v2: > >>> >> - Verify with BIOS version 3.06.25 and 3.07.09 > >>> >> v1: > >>> >> - Initial version > >>> >> --- > >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- > >>> >> 1 files changed, 32 insertions(+), 3 deletions(-) > >>> >> > >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c > >>> >> index f1b633b..00aaa3d 100644 > >>> >> --- a/drivers/pci/host/pci-xgene-msi.c > >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c > >>> >> @@ -24,6 +24,7 @@ > >>> >> #include <linux/pci.h> > >>> >> #include <linux/platform_device.h> > >>> >> #include <linux/of_pci.h> > >>> >> +#include <linux/acpi.h> > >>> >> > >>> >> #define MSI_IR0 0x000000 > >>> >> #define MSI_INT0 0x800000 > >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { > >>> >> }; > >>> >> > >>> >> struct xgene_msi { > >>> >> - struct device_node *node; > >>> >> + struct fwnode_handle *fwnode; > >>> >> struct irq_domain *inner_domain; > >>> >> struct irq_domain *msi_domain; > >>> >> u64 msi_addr; > >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, > >>> >> .free = xgene_irq_domain_free, > >>> >> }; > >>> >> > >>> >> +#ifdef CONFIG_ACPI > >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) > >>> >> +{ > >>> >> + return xgene_msi_ctrl.fwnode; > >>> >> +} > >>> >> +#endif > >>> >> + > >>> >> static int xgene_allocate_domains(struct xgene_msi *msi) > >>> >> { > >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, > >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >>> >> if (!msi->inner_domain) > >>> >> return -ENOMEM; > >>> >> > >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), > >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, > >>> >> &xgene_msi_domain_info, > >>> >> msi->inner_domain); > >>> >> > >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >>> >> return -ENOMEM; > >>> >> } > >>> >> > >>> >> +#ifdef CONFIG_ACPI > >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); > >>> >> +#endif > >>> >> return 0; > >>> >> } > >>> >> > >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) > >>> >> {}, > >>> >> }; > >>> >> > >>> >> +#ifdef CONFIG_ACPI > >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { > >>> >> + {"APMC0D0E", 0}, > >>> >> + { }, > >>> >> +}; > >>> >> +#endif > >>> >> + > >>> >> static int xgene_msi_probe(struct platform_device *pdev) > >>> >> { > >>> >> struct resource *res; > >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) > >>> >> goto error; > >>> >> } > >>> >> xgene_msi->msi_addr = res->start; > >>> >> - xgene_msi->node = pdev->dev.of_node; > >>> >> + > >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); > >>> >> + if (!xgene_msi->fwnode) { > >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); > >>> > > >>> > Please provide something other than NULL, such as the base address if > >>> > the device. That's quite useful for debugging. > >>> > > >>> >> + if (!xgene_msi->fwnode) { > >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); > >>> >> + rc = ENOMEM; > >>> >> + goto error; > >>> >> + } > >>> >> + } > >>> >> + > >>> >> xgene_msi->num_cpus = num_possible_cpus(); > >>> >> > >>> >> rc = xgene_msi_init_allocator(xgene_msi); > >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) > >>> >> .driver = { > >>> >> .name = "xgene-msi", > >>> >> .of_match_table = xgene_msi_match_table, > >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), > >>> >> }, > >>> >> .probe = xgene_msi_probe, > >>> >> .remove = xgene_msi_remove, > >>> >> > >>> > > >>> > The code is trivial, but relies on the MSI controller to probe before > >>> > the PCI bus. What enforces this? How is it making sure that this is not > >>> > going to break in the next kernel release? As far as I can tell, there > >>> > is no explicit dependency between the two, making it the whole thing > >>> > extremely fragile. > >>> > > >>> > Thanks, > >>> > > >>> > M. > >>> > -- > >>> > Jazz is not dead. It just smells funny... > >>> > >>> --
Hi Lorenzo, On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote: > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote: >> Hi Lorenzo, >> >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote: >> > Hi Lorenzo, >> > >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi >> > <lorenzo.pieralisi@arm.com> wrote: >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: >> >>> Hi Marc, >> >>> There's no explicit dependency between pcie driver and msi >> >>> controller. The current solution that we did is relying on the >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a >> >>> higher priority in start ordering. This method could be applied in >> >>> case of msi and pcie are the same level of subsys_init (in ACPI >> >>> boot). However, PCIe driver has not supported for this dependency >> >>> check yet. How do you think about this solution. >> >> >> >> First off, you can't post emails with confidentiality notices on >> >> mailing lists so remove it from now onwards. >> > >> > Fixed >> > >> >> Secondly, I commented on this before, so you know what my opinion is. >> > >> > I got your opinion. I'll implement as your suggestion. >> > >> >> Regarding to your previous suggestion to add a hook walking the ACPI >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI >> controllers must be probed before PCI. I have a concern that the >> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk >> the ACPI namespace and use this name to make MSI probed before PCI. >> May you have little bit more information for this or do you have >> another suggestion for this? >> >> There’s another solution which makes this simpler and I’d like to >> ask your opinion about this. >> The solution is to make an hierarchy between MSI and PCI nodes as below: >> Device(\_SB.MSIX) { >> Device(\_SB.PCI0) >> Device(\_SB.PCI1) >> …… >> } >> In other word, MSIX node is a parent node of PCI nodes in ACPI >> table. In this sense, there’s an explicit dependency between MSI >> and PCI, MSI controller must be probed before PCI and it will >> guarantee not breaking next kernel releases. How do you think about >> this solution. > > I think that's a plaster as ineffective as reordering nodes, in short > it is not a solution and you still rely on kernel link ordering, you > can fiddle with ACPI tables as much as you want but that does not change. > >> I also tried to use _DEP method to describe the dependency of PCIe >> nodes, but it looks that it does not work as PCI and MSI are handed >> by acpi_bus_scan and still having issue when we re-probe PCI. > > That's a tad vague to be frank, anyway it is pretty clear to me that we > have hit a wall. In ACPI there is no way to describe probe dependencies > like the one you have, it is as simple as that, and this MSI issue you > are facing is just an example, there are more eg: > > https://www.spinics.net/lists/arm-kernel/msg585825.html > > At the end of the day the choice is simple either we accept (and we > maintain because that's the problem) these hacks - and I am not willing > to do that - or we find a way to solve this from a general perspective not > as a point hack. > > I can have a look at solving the whole issue but it won't happen > tomorrow. Thanks for helping to resolve this general ACPI dependence issue. I look forward for a version to test with. Can we separate the general dependence patch from the X-Gene MSI ACPI enable patch. This original patch was to enable ACPI support for PCIe MSI. The dependence issue can be resolved separately. Can you help to pull in the patch. Best regards, Khuong Dinh > Thanks, > Lorenzo > >> Thanks, >> Khuong Dinh >> >> >> Finally, please execute this command on the vmlinux that "works" >> >> for you: >> >> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' >> > >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' >> > ffff000008dab2d8 t __initcall_acpi_init4 >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 >> > >> > Best regards, >> > Khuong Dinh >> > >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do >> >> not understand how this works (I mean without relying on kernel link >> >> order to ensure that the MSI driver is probed before PCI devices are >> >> enumerated in acpi_init()). >> >> >> >> Thanks, >> >> Lorenzo >> >> >> >>> Best regards, >> >>> Khuong >> >>> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: >> >>> > On 28/04/17 01:54, Khuong Dinh wrote: >> >>> >> From: Khuong Dinh <kdinh@apm.com> >> >>> >> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >> >>> >> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> >> >>> >> --- >> >>> >> v2: >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09 >> >>> >> v1: >> >>> >> - Initial version >> >>> >> --- >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-) >> >>> >> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c >> >>> >> index f1b633b..00aaa3d 100644 >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c >> >>> >> @@ -24,6 +24,7 @@ >> >>> >> #include <linux/pci.h> >> >>> >> #include <linux/platform_device.h> >> >>> >> #include <linux/of_pci.h> >> >>> >> +#include <linux/acpi.h> >> >>> >> >> >>> >> #define MSI_IR0 0x000000 >> >>> >> #define MSI_INT0 0x800000 >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { >> >>> >> }; >> >>> >> >> >>> >> struct xgene_msi { >> >>> >> - struct device_node *node; >> >>> >> + struct fwnode_handle *fwnode; >> >>> >> struct irq_domain *inner_domain; >> >>> >> struct irq_domain *msi_domain; >> >>> >> u64 msi_addr; >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, >> >>> >> .free = xgene_irq_domain_free, >> >>> >> }; >> >>> >> >> >>> >> +#ifdef CONFIG_ACPI >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) >> >>> >> +{ >> >>> >> + return xgene_msi_ctrl.fwnode; >> >>> >> +} >> >>> >> +#endif >> >>> >> + >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi) >> >>> >> { >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> >>> >> if (!msi->inner_domain) >> >>> >> return -ENOMEM; >> >>> >> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, >> >>> >> &xgene_msi_domain_info, >> >>> >> msi->inner_domain); >> >>> >> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> >>> >> return -ENOMEM; >> >>> >> } >> >>> >> >> >>> >> +#ifdef CONFIG_ACPI >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); >> >>> >> +#endif >> >>> >> return 0; >> >>> >> } >> >>> >> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) >> >>> >> {}, >> >>> >> }; >> >>> >> >> >>> >> +#ifdef CONFIG_ACPI >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { >> >>> >> + {"APMC0D0E", 0}, >> >>> >> + { }, >> >>> >> +}; >> >>> >> +#endif >> >>> >> + >> >>> >> static int xgene_msi_probe(struct platform_device *pdev) >> >>> >> { >> >>> >> struct resource *res; >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) >> >>> >> goto error; >> >>> >> } >> >>> >> xgene_msi->msi_addr = res->start; >> >>> >> - xgene_msi->node = pdev->dev.of_node; >> >>> >> + >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); >> >>> >> + if (!xgene_msi->fwnode) { >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); >> >>> > >> >>> > Please provide something other than NULL, such as the base address if >> >>> > the device. That's quite useful for debugging. >> >>> > >> >>> >> + if (!xgene_msi->fwnode) { >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); >> >>> >> + rc = ENOMEM; >> >>> >> + goto error; >> >>> >> + } >> >>> >> + } >> >>> >> + >> >>> >> xgene_msi->num_cpus = num_possible_cpus(); >> >>> >> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi); >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) >> >>> >> .driver = { >> >>> >> .name = "xgene-msi", >> >>> >> .of_match_table = xgene_msi_match_table, >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), >> >>> >> }, >> >>> >> .probe = xgene_msi_probe, >> >>> >> .remove = xgene_msi_remove, >> >>> >> >> >>> > >> >>> > The code is trivial, but relies on the MSI controller to probe before >> >>> > the PCI bus. What enforces this? How is it making sure that this is not >> >>> > going to break in the next kernel release? As far as I can tell, there >> >>> > is no explicit dependency between the two, making it the whole thing >> >>> > extremely fragile. >> >>> > >> >>> > Thanks, >> >>> > >> >>> > M. >> >>> > -- >> >>> > Jazz is not dead. It just smells funny... >> >>> >> >>> --
On Tue, Jun 13, 2017 at 01:56:44PM -0700, Khuong Dinh wrote: > Hi Lorenzo, > > On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi > <lorenzo.pieralisi@arm.com> wrote: > > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote: > >> Hi Lorenzo, > >> > >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote: > >> > Hi Lorenzo, > >> > > >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi > >> > <lorenzo.pieralisi@arm.com> wrote: > >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: > >> >>> Hi Marc, > >> >>> There's no explicit dependency between pcie driver and msi > >> >>> controller. The current solution that we did is relying on the > >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a > >> >>> higher priority in start ordering. This method could be applied in > >> >>> case of msi and pcie are the same level of subsys_init (in ACPI > >> >>> boot). However, PCIe driver has not supported for this dependency > >> >>> check yet. How do you think about this solution. > >> >> > >> >> First off, you can't post emails with confidentiality notices on > >> >> mailing lists so remove it from now onwards. > >> > > >> > Fixed > >> > > >> >> Secondly, I commented on this before, so you know what my opinion is. > >> > > >> > I got your opinion. I'll implement as your suggestion. > >> > > >> > >> Regarding to your previous suggestion to add a hook walking the ACPI > >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI > >> controllers must be probed before PCI. I have a concern that the > >> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk > >> the ACPI namespace and use this name to make MSI probed before PCI. > >> May you have little bit more information for this or do you have > >> another suggestion for this? > >> > >> There’s another solution which makes this simpler and I’d like to > >> ask your opinion about this. > >> The solution is to make an hierarchy between MSI and PCI nodes as below: > >> Device(\_SB.MSIX) { > >> Device(\_SB.PCI0) > >> Device(\_SB.PCI1) > >> …… > >> } > >> In other word, MSIX node is a parent node of PCI nodes in ACPI > >> table. In this sense, there’s an explicit dependency between MSI > >> and PCI, MSI controller must be probed before PCI and it will > >> guarantee not breaking next kernel releases. How do you think about > >> this solution. > > > > I think that's a plaster as ineffective as reordering nodes, in short > > it is not a solution and you still rely on kernel link ordering, you > > can fiddle with ACPI tables as much as you want but that does not change. > > > >> I also tried to use _DEP method to describe the dependency of PCIe > >> nodes, but it looks that it does not work as PCI and MSI are handed > >> by acpi_bus_scan and still having issue when we re-probe PCI. > > > > That's a tad vague to be frank, anyway it is pretty clear to me that we > > have hit a wall. In ACPI there is no way to describe probe dependencies > > like the one you have, it is as simple as that, and this MSI issue you > > are facing is just an example, there are more eg: > > > > https://www.spinics.net/lists/arm-kernel/msg585825.html > > > > At the end of the day the choice is simple either we accept (and we > > maintain because that's the problem) these hacks - and I am not willing > > to do that - or we find a way to solve this from a general perspective not > > as a point hack. > > > > I can have a look at solving the whole issue but it won't happen > > tomorrow. > > Thanks for helping to resolve this general ACPI dependence issue. > I look forward for a version to test with. > > Can we separate the general dependence patch from the X-Gene MSI ACPI > enable patch. > This original patch was to enable ACPI support for PCIe MSI. > The dependence issue can be resolved separately. > Can you help to pull in the patch. Ok, pragmatically the only sane thing you can do is: (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init()) that carries out fwnode registration (ie register the fwnode by searching the MSI HID object - this does not depend on ACPI device nodes order - you enforce the order by parsing the namespace before ACPI core does it, at that point in time ACPI objects are created). Not sure Rafael will be happy to see this code but that's the only solution that does not rely on ACPI device nodes ordering and kernel link ordering. You may achieve the same by extending the ACPI APD code (drivers/acpi/acpi_apd.c) whether that's the best way forward is to be seen. (2) You can also add an xgene specific scan handler at arch_init_call() but given that PCI root bridge is managed through a scan handler too you would rely on ACPI devices nodes ordering in the DSDT. Let me explain something for the benefit of everyone reading this thread: I do not want X-gene ACPI tables to force the kernel to scan devices in any order whatsover because this would mean we will _never_ be able to change the scan order if we wanted to lest we trigger regressions. So this is a non-option in my opinion. (3) Last option is to register the MSI fwnode either in PCI ECAM quirk handling or in arm64 before probing the PCI root bus. I am not keen on (2) and (3) at all, so _if_ we have to find a solution to this problem (1) is the only option you have for the time being as far as I am concerned. Lorenzo > > Best regards, > Khuong Dinh > > > Thanks, > > Lorenzo > > > >> Thanks, > >> Khuong Dinh > >> > >> >> Finally, please execute this command on the vmlinux that "works" > >> >> for you: > >> >> > >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > >> > > >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > >> > ffff000008dab2d8 t __initcall_acpi_init4 > >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 > >> > > >> > Best regards, > >> > Khuong Dinh > >> > > >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do > >> >> not understand how this works (I mean without relying on kernel link > >> >> order to ensure that the MSI driver is probed before PCI devices are > >> >> enumerated in acpi_init()). > >> >> > >> >> Thanks, > >> >> Lorenzo > >> >> > >> >>> Best regards, > >> >>> Khuong > >> >>> > >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: > >> >>> > On 28/04/17 01:54, Khuong Dinh wrote: > >> >>> >> From: Khuong Dinh <kdinh@apm.com> > >> >>> >> > >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides > >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > >> >>> >> > >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> > >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com> > >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> > >> >>> >> --- > >> >>> >> v2: > >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09 > >> >>> >> v1: > >> >>> >> - Initial version > >> >>> >> --- > >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- > >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-) > >> >>> >> > >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c > >> >>> >> index f1b633b..00aaa3d 100644 > >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c > >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c > >> >>> >> @@ -24,6 +24,7 @@ > >> >>> >> #include <linux/pci.h> > >> >>> >> #include <linux/platform_device.h> > >> >>> >> #include <linux/of_pci.h> > >> >>> >> +#include <linux/acpi.h> > >> >>> >> > >> >>> >> #define MSI_IR0 0x000000 > >> >>> >> #define MSI_INT0 0x800000 > >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { > >> >>> >> }; > >> >>> >> > >> >>> >> struct xgene_msi { > >> >>> >> - struct device_node *node; > >> >>> >> + struct fwnode_handle *fwnode; > >> >>> >> struct irq_domain *inner_domain; > >> >>> >> struct irq_domain *msi_domain; > >> >>> >> u64 msi_addr; > >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, > >> >>> >> .free = xgene_irq_domain_free, > >> >>> >> }; > >> >>> >> > >> >>> >> +#ifdef CONFIG_ACPI > >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) > >> >>> >> +{ > >> >>> >> + return xgene_msi_ctrl.fwnode; > >> >>> >> +} > >> >>> >> +#endif > >> >>> >> + > >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi) > >> >>> >> { > >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, > >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >> >>> >> if (!msi->inner_domain) > >> >>> >> return -ENOMEM; > >> >>> >> > >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), > >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, > >> >>> >> &xgene_msi_domain_info, > >> >>> >> msi->inner_domain); > >> >>> >> > >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >> >>> >> return -ENOMEM; > >> >>> >> } > >> >>> >> > >> >>> >> +#ifdef CONFIG_ACPI > >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); > >> >>> >> +#endif > >> >>> >> return 0; > >> >>> >> } > >> >>> >> > >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) > >> >>> >> {}, > >> >>> >> }; > >> >>> >> > >> >>> >> +#ifdef CONFIG_ACPI > >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { > >> >>> >> + {"APMC0D0E", 0}, > >> >>> >> + { }, > >> >>> >> +}; > >> >>> >> +#endif > >> >>> >> + > >> >>> >> static int xgene_msi_probe(struct platform_device *pdev) > >> >>> >> { > >> >>> >> struct resource *res; > >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) > >> >>> >> goto error; > >> >>> >> } > >> >>> >> xgene_msi->msi_addr = res->start; > >> >>> >> - xgene_msi->node = pdev->dev.of_node; > >> >>> >> + > >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); > >> >>> >> + if (!xgene_msi->fwnode) { > >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); > >> >>> > > >> >>> > Please provide something other than NULL, such as the base address if > >> >>> > the device. That's quite useful for debugging. > >> >>> > > >> >>> >> + if (!xgene_msi->fwnode) { > >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); > >> >>> >> + rc = ENOMEM; > >> >>> >> + goto error; > >> >>> >> + } > >> >>> >> + } > >> >>> >> + > >> >>> >> xgene_msi->num_cpus = num_possible_cpus(); > >> >>> >> > >> >>> >> rc = xgene_msi_init_allocator(xgene_msi); > >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) > >> >>> >> .driver = { > >> >>> >> .name = "xgene-msi", > >> >>> >> .of_match_table = xgene_msi_match_table, > >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), > >> >>> >> }, > >> >>> >> .probe = xgene_msi_probe, > >> >>> >> .remove = xgene_msi_remove, > >> >>> >> > >> >>> > > >> >>> > The code is trivial, but relies on the MSI controller to probe before > >> >>> > the PCI bus. What enforces this? How is it making sure that this is not > >> >>> > going to break in the next kernel release? As far as I can tell, there > >> >>> > is no explicit dependency between the two, making it the whole thing > >> >>> > extremely fragile. > >> >>> > > >> >>> > Thanks, > >> >>> > > >> >>> > M. > >> >>> > -- > >> >>> > Jazz is not dead. It just smells funny... > >> >>> > >> >>> --
On Wed, Jun 14, 2017 at 5:59 AM, Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote: > On Tue, Jun 13, 2017 at 01:56:44PM -0700, Khuong Dinh wrote: >> Hi Lorenzo, >> >> On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi >> <lorenzo.pieralisi@arm.com> wrote: >> > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote: >> >> Hi Lorenzo, >> >> >> >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote: >> >> > Hi Lorenzo, >> >> > >> >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi >> >> > <lorenzo.pieralisi@arm.com> wrote: >> >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote: >> >> >>> Hi Marc, >> >> >>> There's no explicit dependency between pcie driver and msi >> >> >>> controller. The current solution that we did is relying on the >> >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a >> >> >>> higher priority in start ordering. This method could be applied in >> >> >>> case of msi and pcie are the same level of subsys_init (in ACPI >> >> >>> boot). However, PCIe driver has not supported for this dependency >> >> >>> check yet. How do you think about this solution. >> >> >> >> >> >> First off, you can't post emails with confidentiality notices on >> >> >> mailing lists so remove it from now onwards. >> >> > >> >> > Fixed >> >> > >> >> >> Secondly, I commented on this before, so you know what my opinion is. >> >> > >> >> > I got your opinion. I'll implement as your suggestion. >> >> > >> >> >> >> Regarding to your previous suggestion to add a hook walking the ACPI >> >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI >> >> controllers must be probed before PCI. I have a concern that the >> >> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk >> >> the ACPI namespace and use this name to make MSI probed before PCI. >> >> May you have little bit more information for this or do you have >> >> another suggestion for this? >> >> >> >> There’s another solution which makes this simpler and I’d like to >> >> ask your opinion about this. >> >> The solution is to make an hierarchy between MSI and PCI nodes as below: >> >> Device(\_SB.MSIX) { >> >> Device(\_SB.PCI0) >> >> Device(\_SB.PCI1) >> >> …… >> >> } >> >> In other word, MSIX node is a parent node of PCI nodes in ACPI >> >> table. In this sense, there’s an explicit dependency between MSI >> >> and PCI, MSI controller must be probed before PCI and it will >> >> guarantee not breaking next kernel releases. How do you think about >> >> this solution. >> > >> > I think that's a plaster as ineffective as reordering nodes, in short >> > it is not a solution and you still rely on kernel link ordering, you >> > can fiddle with ACPI tables as much as you want but that does not change. >> > >> >> I also tried to use _DEP method to describe the dependency of PCIe >> >> nodes, but it looks that it does not work as PCI and MSI are handed >> >> by acpi_bus_scan and still having issue when we re-probe PCI. >> > >> > That's a tad vague to be frank, anyway it is pretty clear to me that we >> > have hit a wall. In ACPI there is no way to describe probe dependencies >> > like the one you have, it is as simple as that, and this MSI issue you >> > are facing is just an example, there are more eg: >> > >> > https://www.spinics.net/lists/arm-kernel/msg585825.html >> > >> > At the end of the day the choice is simple either we accept (and we >> > maintain because that's the problem) these hacks - and I am not willing >> > to do that - or we find a way to solve this from a general perspective not >> > as a point hack. >> > >> > I can have a look at solving the whole issue but it won't happen >> > tomorrow. >> >> Thanks for helping to resolve this general ACPI dependence issue. >> I look forward for a version to test with. >> >> Can we separate the general dependence patch from the X-Gene MSI ACPI >> enable patch. >> This original patch was to enable ACPI support for PCIe MSI. >> The dependence issue can be resolved separately. >> Can you help to pull in the patch. > > Ok, pragmatically the only sane thing you can do is: > > (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init()) > that carries out fwnode registration (ie register the fwnode by > searching the MSI HID object - this does not depend on ACPI device > nodes order - you enforce the order by parsing the namespace before > ACPI core does it, at that point in time ACPI objects are created). > Not sure Rafael will be happy to see this code but that's the only > solution that does not rely on ACPI device nodes ordering and kernel > link ordering. You may achieve the same by extending the ACPI APD > code (drivers/acpi/acpi_apd.c) whether that's the best way forward is > to be seen. Hi Rafael, Do you have any idea about this suggestion? Otherwise, I'll follow this Lorenzo's approach (1). > (2) You can also add an xgene specific scan handler at arch_init_call() > but given that PCI root bridge is managed through a scan handler too you > would rely on ACPI devices nodes ordering in the DSDT. Let me explain > something for the benefit of everyone reading this thread: I do not want > X-gene ACPI tables to force the kernel to scan devices in any order > whatsover because this would mean we will _never_ be able to change the > scan order if we wanted to lest we trigger regressions. So this is > a non-option in my opinion. > > (3) Last option is to register the MSI fwnode either in PCI ECAM quirk > handling or in arm64 before probing the PCI root bus. > > I am not keen on (2) and (3) at all, so _if_ we have to find a solution > to this problem (1) is the only option you have for the time being as > far as I am concerned. Thank you very much for your suggestions, Lorenzo. If there's not any other idea, I'll follow your approach (1) Best regards, Khuong Dinh > Lorenzo > >> >> Best regards, >> Khuong Dinh >> >> > Thanks, >> > Lorenzo >> > >> >> Thanks, >> >> Khuong Dinh >> >> >> >> >> Finally, please execute this command on the vmlinux that "works" >> >> >> for you: >> >> >> >> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' >> >> > >> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' >> >> > ffff000008dab2d8 t __initcall_acpi_init4 >> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 >> >> > >> >> > Best regards, >> >> > Khuong Dinh >> >> > >> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do >> >> >> not understand how this works (I mean without relying on kernel link >> >> >> order to ensure that the MSI driver is probed before PCI devices are >> >> >> enumerated in acpi_init()). >> >> >> >> >> >> Thanks, >> >> >> Lorenzo >> >> >> >> >> >>> Best regards, >> >> >>> Khuong >> >> >>> >> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: >> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote: >> >> >>> >> From: Khuong Dinh <kdinh@apm.com> >> >> >>> >> >> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides >> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. >> >> >>> >> >> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> >> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com> >> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> >> >> >>> >> --- >> >> >>> >> v2: >> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09 >> >> >>> >> v1: >> >> >>> >> - Initial version >> >> >>> >> --- >> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- >> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-) >> >> >>> >> >> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c >> >> >>> >> index f1b633b..00aaa3d 100644 >> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c >> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c >> >> >>> >> @@ -24,6 +24,7 @@ >> >> >>> >> #include <linux/pci.h> >> >> >>> >> #include <linux/platform_device.h> >> >> >>> >> #include <linux/of_pci.h> >> >> >>> >> +#include <linux/acpi.h> >> >> >>> >> >> >> >>> >> #define MSI_IR0 0x000000 >> >> >>> >> #define MSI_INT0 0x800000 >> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { >> >> >>> >> }; >> >> >>> >> >> >> >>> >> struct xgene_msi { >> >> >>> >> - struct device_node *node; >> >> >>> >> + struct fwnode_handle *fwnode; >> >> >>> >> struct irq_domain *inner_domain; >> >> >>> >> struct irq_domain *msi_domain; >> >> >>> >> u64 msi_addr; >> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, >> >> >>> >> .free = xgene_irq_domain_free, >> >> >>> >> }; >> >> >>> >> >> >> >>> >> +#ifdef CONFIG_ACPI >> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) >> >> >>> >> +{ >> >> >>> >> + return xgene_msi_ctrl.fwnode; >> >> >>> >> +} >> >> >>> >> +#endif >> >> >>> >> + >> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi) >> >> >>> >> { >> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, >> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> >> >>> >> if (!msi->inner_domain) >> >> >>> >> return -ENOMEM; >> >> >>> >> >> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), >> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, >> >> >>> >> &xgene_msi_domain_info, >> >> >>> >> msi->inner_domain); >> >> >>> >> >> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) >> >> >>> >> return -ENOMEM; >> >> >>> >> } >> >> >>> >> >> >> >>> >> +#ifdef CONFIG_ACPI >> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); >> >> >>> >> +#endif >> >> >>> >> return 0; >> >> >>> >> } >> >> >>> >> >> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) >> >> >>> >> {}, >> >> >>> >> }; >> >> >>> >> >> >> >>> >> +#ifdef CONFIG_ACPI >> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { >> >> >>> >> + {"APMC0D0E", 0}, >> >> >>> >> + { }, >> >> >>> >> +}; >> >> >>> >> +#endif >> >> >>> >> + >> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev) >> >> >>> >> { >> >> >>> >> struct resource *res; >> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) >> >> >>> >> goto error; >> >> >>> >> } >> >> >>> >> xgene_msi->msi_addr = res->start; >> >> >>> >> - xgene_msi->node = pdev->dev.of_node; >> >> >>> >> + >> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); >> >> >>> >> + if (!xgene_msi->fwnode) { >> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); >> >> >>> > >> >> >>> > Please provide something other than NULL, such as the base address if >> >> >>> > the device. That's quite useful for debugging. >> >> >>> > >> >> >>> >> + if (!xgene_msi->fwnode) { >> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); >> >> >>> >> + rc = ENOMEM; >> >> >>> >> + goto error; >> >> >>> >> + } >> >> >>> >> + } >> >> >>> >> + >> >> >>> >> xgene_msi->num_cpus = num_possible_cpus(); >> >> >>> >> >> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi); >> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) >> >> >>> >> .driver = { >> >> >>> >> .name = "xgene-msi", >> >> >>> >> .of_match_table = xgene_msi_match_table, >> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), >> >> >>> >> }, >> >> >>> >> .probe = xgene_msi_probe, >> >> >>> >> .remove = xgene_msi_remove, >> >> >>> >> >> >> >>> > >> >> >>> > The code is trivial, but relies on the MSI controller to probe before >> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not >> >> >>> > going to break in the next kernel release? As far as I can tell, there >> >> >>> > is no explicit dependency between the two, making it the whole thing >> >> >>> > extremely fragile. >> >> >>> > >> >> >>> > Thanks, >> >> >>> > >> >> >>> > M. >> >> >>> > -- >> >> >>> > Jazz is not dead. It just smells funny... >> >> >>> >> >> >>> --
On Wed, Jun 14, 2017 at 10:43:23AM -0700, Khuong Dinh wrote: [...] > >> Thanks for helping to resolve this general ACPI dependence issue. > >> I look forward for a version to test with. > >> > >> Can we separate the general dependence patch from the X-Gene MSI ACPI > >> enable patch. > >> This original patch was to enable ACPI support for PCIe MSI. > >> The dependence issue can be resolved separately. > >> Can you help to pull in the patch. > > > > Ok, pragmatically the only sane thing you can do is: > > > > (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init()) > > that carries out fwnode registration (ie register the fwnode by > > searching the MSI HID object - this does not depend on ACPI device > > nodes order - you enforce the order by parsing the namespace before > > ACPI core does it, at that point in time ACPI objects are created). > > Not sure Rafael will be happy to see this code but that's the only > > solution that does not rely on ACPI device nodes ordering and kernel > > link ordering. You may achieve the same by extending the ACPI APD > > code (drivers/acpi/acpi_apd.c) whether that's the best way forward is > > to be seen. > > Hi Rafael, > Do you have any idea about this suggestion? I mentioned drivers/acpi/acpi_apd.c because is in a way "similar" to what you need but it is not exactly what's needed (I am not sure what "devices" acpi_apd.c is meant to manage at the moment but certainly not MSI controllers). In your callback from ACPI code you can't add a scan handler - you have to get the ACPI device corresponding to the MSI controller and register the IRQ domain fwnode. Honestly it is hackish but that's the only approach that is robust enough to always work (ie it does not depend on ACPI tables nodes order and it does not depend on kernel link order). Lorenzo > Otherwise, I'll follow this Lorenzo's approach (1). > > > (2) You can also add an xgene specific scan handler at arch_init_call() > > but given that PCI root bridge is managed through a scan handler too you > > would rely on ACPI devices nodes ordering in the DSDT. Let me explain > > something for the benefit of everyone reading this thread: I do not want > > X-gene ACPI tables to force the kernel to scan devices in any order > > whatsover because this would mean we will _never_ be able to change the > > scan order if we wanted to lest we trigger regressions. So this is > > a non-option in my opinion. > > > > (3) Last option is to register the MSI fwnode either in PCI ECAM quirk > > handling or in arm64 before probing the PCI root bus. > > > > I am not keen on (2) and (3) at all, so _if_ we have to find a solution > > to this problem (1) is the only option you have for the time being as > > far as I am concerned. > > Thank you very much for your suggestions, Lorenzo. > If there's not any other idea, I'll follow your approach (1) > > Best regards, > Khuong Dinh > > > Lorenzo > > > >> > >> Best regards, > >> Khuong Dinh > >> > >> > Thanks, > >> > Lorenzo > >> > > >> >> Thanks, > >> >> Khuong Dinh > >> >> > >> >> >> Finally, please execute this command on the vmlinux that "works" > >> >> >> for you: > >> >> >> > >> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > >> >> > > >> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)' > >> >> > ffff000008dab2d8 t __initcall_acpi_init4 > >> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4 > >> >> > > >> >> > Best regards, > >> >> > Khuong Dinh > >> >> > > >> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do > >> >> >> not understand how this works (I mean without relying on kernel link > >> >> >> order to ensure that the MSI driver is probed before PCI devices are > >> >> >> enumerated in acpi_init()). > >> >> >> > >> >> >> Thanks, > >> >> >> Lorenzo > >> >> >> > >> >> >>> Best regards, > >> >> >>> Khuong > >> >> >>> > >> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote: > >> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote: > >> >> >>> >> From: Khuong Dinh <kdinh@apm.com> > >> >> >>> >> > >> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides > >> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode. > >> >> >>> >> > >> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com> > >> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com> > >> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com> > >> >> >>> >> --- > >> >> >>> >> v2: > >> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09 > >> >> >>> >> v1: > >> >> >>> >> - Initial version > >> >> >>> >> --- > >> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++--- > >> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-) > >> >> >>> >> > >> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c > >> >> >>> >> index f1b633b..00aaa3d 100644 > >> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c > >> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c > >> >> >>> >> @@ -24,6 +24,7 @@ > >> >> >>> >> #include <linux/pci.h> > >> >> >>> >> #include <linux/platform_device.h> > >> >> >>> >> #include <linux/of_pci.h> > >> >> >>> >> +#include <linux/acpi.h> > >> >> >>> >> > >> >> >>> >> #define MSI_IR0 0x000000 > >> >> >>> >> #define MSI_INT0 0x800000 > >> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group { > >> >> >>> >> }; > >> >> >>> >> > >> >> >>> >> struct xgene_msi { > >> >> >>> >> - struct device_node *node; > >> >> >>> >> + struct fwnode_handle *fwnode; > >> >> >>> >> struct irq_domain *inner_domain; > >> >> >>> >> struct irq_domain *msi_domain; > >> >> >>> >> u64 msi_addr; > >> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, > >> >> >>> >> .free = xgene_irq_domain_free, > >> >> >>> >> }; > >> >> >>> >> > >> >> >>> >> +#ifdef CONFIG_ACPI > >> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) > >> >> >>> >> +{ > >> >> >>> >> + return xgene_msi_ctrl.fwnode; > >> >> >>> >> +} > >> >> >>> >> +#endif > >> >> >>> >> + > >> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi) > >> >> >>> >> { > >> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, > >> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >> >> >>> >> if (!msi->inner_domain) > >> >> >>> >> return -ENOMEM; > >> >> >>> >> > >> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), > >> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, > >> >> >>> >> &xgene_msi_domain_info, > >> >> >>> >> msi->inner_domain); > >> >> >>> >> > >> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) > >> >> >>> >> return -ENOMEM; > >> >> >>> >> } > >> >> >>> >> > >> >> >>> >> +#ifdef CONFIG_ACPI > >> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); > >> >> >>> >> +#endif > >> >> >>> >> return 0; > >> >> >>> >> } > >> >> >>> >> > >> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) > >> >> >>> >> {}, > >> >> >>> >> }; > >> >> >>> >> > >> >> >>> >> +#ifdef CONFIG_ACPI > >> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = { > >> >> >>> >> + {"APMC0D0E", 0}, > >> >> >>> >> + { }, > >> >> >>> >> +}; > >> >> >>> >> +#endif > >> >> >>> >> + > >> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev) > >> >> >>> >> { > >> >> >>> >> struct resource *res; > >> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) > >> >> >>> >> goto error; > >> >> >>> >> } > >> >> >>> >> xgene_msi->msi_addr = res->start; > >> >> >>> >> - xgene_msi->node = pdev->dev.of_node; > >> >> >>> >> + > >> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); > >> >> >>> >> + if (!xgene_msi->fwnode) { > >> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); > >> >> >>> > > >> >> >>> > Please provide something other than NULL, such as the base address if > >> >> >>> > the device. That's quite useful for debugging. > >> >> >>> > > >> >> >>> >> + if (!xgene_msi->fwnode) { > >> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n"); > >> >> >>> >> + rc = ENOMEM; > >> >> >>> >> + goto error; > >> >> >>> >> + } > >> >> >>> >> + } > >> >> >>> >> + > >> >> >>> >> xgene_msi->num_cpus = num_possible_cpus(); > >> >> >>> >> > >> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi); > >> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) > >> >> >>> >> .driver = { > >> >> >>> >> .name = "xgene-msi", > >> >> >>> >> .of_match_table = xgene_msi_match_table, > >> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), > >> >> >>> >> }, > >> >> >>> >> .probe = xgene_msi_probe, > >> >> >>> >> .remove = xgene_msi_remove, > >> >> >>> >> > >> >> >>> > > >> >> >>> > The code is trivial, but relies on the MSI controller to probe before > >> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not > >> >> >>> > going to break in the next kernel release? As far as I can tell, there > >> >> >>> > is no explicit dependency between the two, making it the whole thing > >> >> >>> > extremely fragile. > >> >> >>> > > >> >> >>> > Thanks, > >> >> >>> > > >> >> >>> > M. > >> >> >>> > -- > >> >> >>> > Jazz is not dead. It just smells funny... > >> >> >>> > >> >> >>> --
diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c index f1b633b..00aaa3d 100644 --- a/drivers/pci/host/pci-xgene-msi.c +++ b/drivers/pci/host/pci-xgene-msi.c @@ -24,6 +24,7 @@ #include <linux/pci.h> #include <linux/platform_device.h> #include <linux/of_pci.h> +#include <linux/acpi.h> #define MSI_IR0 0x000000 #define MSI_INT0 0x800000 @@ -39,7 +40,7 @@ struct xgene_msi_group { }; struct xgene_msi { - struct device_node *node; + struct fwnode_handle *fwnode; struct irq_domain *inner_domain; struct irq_domain *msi_domain; u64 msi_addr; @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain, .free = xgene_irq_domain_free, }; +#ifdef CONFIG_ACPI +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev) +{ + return xgene_msi_ctrl.fwnode; +} +#endif + static int xgene_allocate_domains(struct xgene_msi *msi) { msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC, @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi) if (!msi->inner_domain) return -ENOMEM; - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node), + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode, &xgene_msi_domain_info, msi->inner_domain); @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi) return -ENOMEM; } +#ifdef CONFIG_ACPI + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode); +#endif return 0; } @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu) {}, }; +#ifdef CONFIG_ACPI +static const struct acpi_device_id xgene_msi_acpi_ids[] = { + {"APMC0D0E", 0}, + { }, +}; +#endif + static int xgene_msi_probe(struct platform_device *pdev) { struct resource *res; @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev) goto error; } xgene_msi->msi_addr = res->start; - xgene_msi->node = pdev->dev.of_node; + + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node); + if (!xgene_msi->fwnode) { + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL); + if (!xgene_msi->fwnode) { + dev_err(&pdev->dev, "Failed to create fwnode\n"); + rc = ENOMEM; + goto error; + } + } + xgene_msi->num_cpus = num_possible_cpus(); rc = xgene_msi_init_allocator(xgene_msi); @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev) .driver = { .name = "xgene-msi", .of_match_table = xgene_msi_match_table, + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids), }, .probe = xgene_msi_probe, .remove = xgene_msi_remove,