From patchwork Fri Jan 23 04:29:36 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Beth Kon X-Patchwork-Id: 3722 Received: from vger.kernel.org (vger.kernel.org [209.132.176.167]) by demeter.kernel.org (8.14.2/8.14.2) with ESMTP id n0N4T168028293 for ; Thu, 22 Jan 2009 20:29:02 -0800 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755596AbZAWEdl (ORCPT ); Thu, 22 Jan 2009 23:33:41 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755484AbZAWEdl (ORCPT ); Thu, 22 Jan 2009 23:33:41 -0500 Received: from e5.ny.us.ibm.com ([32.97.182.145]:34150 "EHLO e5.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755482AbZAWEdj (ORCPT ); Thu, 22 Jan 2009 23:33:39 -0500 Received: from d01relay04.pok.ibm.com (d01relay04.pok.ibm.com [9.56.227.236]) by e5.ny.us.ibm.com (8.13.1/8.13.1) with ESMTP id n0N4Vj2G024303 for ; Thu, 22 Jan 2009 23:31:45 -0500 Received: from d01av01.pok.ibm.com (d01av01.pok.ibm.com [9.56.224.215]) by d01relay04.pok.ibm.com (8.13.8/8.13.8/NCO v9.1) with ESMTP id n0N4Xcg7191662 for ; Thu, 22 Jan 2009 23:33:38 -0500 Received: from d01av01.pok.ibm.com (loopback [127.0.0.1]) by d01av01.pok.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id n0N4Xccr004574 for ; Thu, 22 Jan 2009 23:33:38 -0500 Received: from localhost.localdomain (npt1.ltc.austin.ibm.com [9.3.110.129]) by d01av01.pok.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id n0N4XbSo004569; Thu, 22 Jan 2009 23:33:37 -0500 From: Beth Kon To: kvm@vger.kernel.org Cc: Beth Kon Subject: [RFC][PATCH 1/2] Make irq0->inti2 override in BIOS configurable from userspace Date: Thu, 22 Jan 2009 22:29:36 -0600 Message-Id: <1232684977-4530-1-git-send-email-eak@us.ibm.com> X-Mailer: git-send-email 1.5.4.3 Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org This series of patches (nearly) resolves the irq0->inti2 override issue, and gets the hpet working on kvm with and without the in-kernel irqchip (i.e., it disables userspace and in-kernel pit as needed). - irq0->inti2 The resolution was to always use the override unless the kernel cannot do irq routing (i.e., compatibility with old kernels). So qemu checks whether the kernel is capable of irq routing. If so, qemu tells kvm to route irq0 to inti2 via the irq routing interface, and tells bios to add the irq0->inti2 override to the MADT interrupt source override table, and to the mp table (for the non-acpi case). The only outstanding problem here is that when I set acpi=off on the kernel boot line, the boot fails. Apparently linux does not like the way I implemented the override for the mp table in rombios32.c. Since I am pressed for time at the moment, I'm putting this patch set out for comments in hopes that someone else may immediately see the problem. Otherwise I'll keep looking into it as time permits. - hpet The hpet works with and without in-kernel irqchip. And many thanks to Marcelo for finding a bios corruption bug that was the primary source of win2k864 problems. Now the hpet works on linux (ubuntu 8.0.4), win2k832. On win2k864 it works with the in-kernel irqchip but is broken (i.e.,black screen) when -no-kvm-irqchip is specified. Though I found that it is also broken when I remove these 2 patches, so it appears to have nothing to do with hpet or irq routing. Needs more looking into. Signed-off-by: Beth Kon --- bios/Makefile | 2 +- bios/rombios32.c | 40 ++++++++++++++++++++++++++++++++++++---- qemu/hw/apic.c | 5 ++--- qemu/hw/fw_cfg.c | 1 + qemu/hw/fw_cfg.h | 1 + qemu/qemu-kvm.c | 5 ++++- qemu/sysemu.h | 1 + qemu/vl.c | 10 ++++++++-- 8 files changed, 54 insertions(+), 11 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/bios/Makefile b/bios/Makefile index 2d1f40d..374d70e 100644 --- a/bios/Makefile +++ b/bios/Makefile @@ -48,7 +48,7 @@ LIBS = -lm RANLIB = ranlib BCC = bcc -GCC = gcc $(CFLAGS) +GCC = gcc $(CFLAGS) -fno-stack-protector HOST_CC = gcc AS86 = as86 diff --git a/bios/rombios32.c b/bios/rombios32.c index 9d2eaaa..84f15fb 100755 --- a/bios/rombios32.c +++ b/bios/rombios32.c @@ -443,6 +443,7 @@ uint32_t cpuid_ext_features; unsigned long ram_size; uint64_t ram_end; uint8_t bios_uuid[16]; +uint8_t irq0_override; #ifdef BX_USE_EBDA_TABLES unsigned long ebda_cur_addr; #endif @@ -475,6 +476,7 @@ void wrmsr_smp(uint32_t index, uint64_t val) #define QEMU_CFG_SIGNATURE 0x00 #define QEMU_CFG_ID 0x01 #define QEMU_CFG_UUID 0x02 +#define QEMU_CFG_IRQ0_OVERRIDE 0x07 int qemu_cfg_port; @@ -516,6 +518,18 @@ void uuid_probe(void) memset(bios_uuid, 0, 16); } +void irq0_override_probe(void) +{ +#ifdef BX_QEMU + if(qemu_cfg_port) { + qemu_cfg_select(QEMU_CFG_IRQ0_OVERRIDE); + qemu_cfg_read(&irq0_override, 1); + return; + } +#endif + memset(&irq0_override, 0, 1); +} + void cpu_probe(void) { uint32_t eax, ebx, ecx, edx; @@ -1149,6 +1163,8 @@ static void mptable_init(void) /* irqs */ for(i = 0; i < 16; i++) { + if (irq0_override && i == 2) + continue; putb(&q, 3); /* entry type = I/O interrupt */ putb(&q, 0); /* interrupt type = vectored interrupt */ putb(&q, 0); /* flags: po=0, el=0 */ @@ -1156,7 +1172,10 @@ static void mptable_init(void) putb(&q, 0); /* source bus ID = ISA */ putb(&q, i); /* source bus IRQ */ putb(&q, ioapic_id); /* dest I/O APIC ID */ - putb(&q, i); /* dest I/O APIC interrupt in */ + if (irq0_override && i == 0) + putb(&q, 2); /* dest I/O APIC interrupt in */ + else + putb(&q, i); /* dest I/O APIC interrupt in */ } /* patch length */ len = q - mp_config_table; @@ -1505,6 +1524,11 @@ void acpi_bios_init(void) sizeof(struct madt_processor_apic) * MAX_CPUS + sizeof(struct madt_io_apic); madt = (void *)(addr); + for (i = 0; i < 16; i++) + if (PCI_ISA_IRQ_MASK & (1U << i)) + madt_size += sizeof(struct madt_intsrcovr); + if (irq0_override) + madt_size += sizeof(struct madt_intsrcovr); addr += madt_size; acpi_tables_size = addr - base_addr; @@ -1594,8 +1618,15 @@ void acpi_bios_init(void) io_apic->interrupt = cpu_to_le32(0); intsrcovr = (struct madt_intsrcovr*)(io_apic + 1); - for ( i = 0; i < 16; i++ ) { - if ( PCI_ISA_IRQ_MASK & (1U << i) ) { + for (i = 0; i < 16; i++) { + if (irq0_override && i == 0) { + memset(intsrcovr, 0, sizeof(*intsrcovr)); + intsrcovr->type = APIC_XRUPT_OVERRIDE; + intsrcovr->length = sizeof(*intsrcovr); + intsrcovr->source = i; + intsrcovr->gsi = 2; + intsrcovr->flags = 0; //conforms to bus specifications + } else if (PCI_ISA_IRQ_MASK & (1U << i)) { memset(intsrcovr, 0, sizeof(*intsrcovr)); intsrcovr->type = APIC_XRUPT_OVERRIDE; intsrcovr->length = sizeof(*intsrcovr); @@ -1607,7 +1638,6 @@ void acpi_bios_init(void) continue; } intsrcovr++; - madt_size += sizeof(struct madt_intsrcovr); } acpi_build_table_header((struct acpi_table_header *)madt, "APIC", madt_size, 1); @@ -2232,6 +2262,8 @@ void rombios32_init(uint32_t *s3_resume_vector, uint8_t *shutdown_flag) uuid_probe(); smbios_init(); + + irq0_override_probe(); if (acpi_enabled) acpi_bios_init(); diff --git a/qemu/hw/apic.c b/qemu/hw/apic.c index 782b398..f28a103 100644 --- a/qemu/hw/apic.c +++ b/qemu/hw/apic.c @@ -19,6 +19,7 @@ */ #include "hw.h" #include "pc.h" +#include "sysemu.h" #include "qemu-timer.h" #include "host-utils.h" @@ -1077,14 +1078,12 @@ void ioapic_set_irq(void *opaque, int vector, int level) { IOAPICState *s = opaque; -#if 0 /* ISA IRQs map to GSI 1-1 except for IRQ0 which maps * to GSI 2. GSI maps to ioapic 1-1. This is not * the cleanest way of doing it but it should work. */ - if (vector == 0) + if (vector == 0 && irq0override) vector = 2; -#endif if (vector >= 0 && vector < IOAPIC_NUM_PINS) { uint32_t mask = 1 << vector; diff --git a/qemu/hw/fw_cfg.c b/qemu/hw/fw_cfg.c index 4333ed9..1b31ad2 100644 --- a/qemu/hw/fw_cfg.c +++ b/qemu/hw/fw_cfg.c @@ -287,6 +287,7 @@ void *fw_cfg_init(uint32_t ctl_port, uint32_t data_port, fw_cfg_add_bytes(s, FW_CFG_UUID, qemu_uuid, 16); fw_cfg_add_i16(s, FW_CFG_NOGRAPHIC, (uint16_t)nographic); fw_cfg_add_i16(s, FW_CFG_NB_CPUS, (uint16_t)smp_cpus); + fw_cfg_add_i16(s, FW_CFG_IRQ0_OVERRIDE, (uint16_t)irq0override); register_savevm("fw_cfg", -1, 1, fw_cfg_save, fw_cfg_load, s); qemu_register_reset(fw_cfg_reset, s); diff --git a/qemu/hw/fw_cfg.h b/qemu/hw/fw_cfg.h index ef8f378..a4e81d2 100644 --- a/qemu/hw/fw_cfg.h +++ b/qemu/hw/fw_cfg.h @@ -8,6 +8,7 @@ #define FW_CFG_NOGRAPHIC 0x04 #define FW_CFG_NB_CPUS 0x05 #define FW_CFG_MACHINE_ID 0x06 +#define FW_CFG_IRQ0_OVERRIDE 0x07 #define FW_CFG_MAX_ENTRY 0x10 #define FW_CFG_WRITE_CHANNEL 0x4000 diff --git a/qemu/qemu-kvm.c b/qemu/qemu-kvm.c index 5ff63ad..e32b4c2 100644 --- a/qemu/qemu-kvm.c +++ b/qemu/qemu-kvm.c @@ -815,7 +815,10 @@ int kvm_qemu_create_context(void) return r; } for (i = 0; i < 24; ++i) { - r = kvm_add_irq_route(kvm_context, i, KVM_IRQCHIP_IOAPIC, i); + if (i == 0) + r = kvm_add_irq_route(kvm_context, i, KVM_IRQCHIP_IOAPIC, 2); + else + r = kvm_add_irq_route(kvm_context, i, KVM_IRQCHIP_IOAPIC, i); if (r < 0) return r; } diff --git a/qemu/sysemu.h b/qemu/sysemu.h index c8f268e..04db217 100644 --- a/qemu/sysemu.h +++ b/qemu/sysemu.h @@ -92,6 +92,7 @@ extern int graphic_width; extern int graphic_height; extern int graphic_depth; extern int nographic; +extern int irq0override; extern const char *keyboard_layout; extern int win2k_install_hack; extern int rtc_td_hack; diff --git a/qemu/vl.c b/qemu/vl.c index 9557a06..21b3e51 100644 --- a/qemu/vl.c +++ b/qemu/vl.c @@ -194,6 +194,7 @@ static int vga_ram_size; enum vga_retrace_method vga_retrace_method = VGA_RETRACE_DUMB; static DisplayState *display_state; int nographic; +int irq0override; static int curses; static int sdl; const char* keyboard_layout = NULL; @@ -4874,6 +4875,7 @@ int main(int argc, char **argv, char **envp) #endif snapshot = 0; nographic = 0; + irq0override = 1; curses = 0; kernel_filename = NULL; kernel_cmdline = ""; @@ -5871,8 +5873,12 @@ int main(int argc, char **argv, char **envp) } } - if (kvm_enabled()) - kvm_init_ap(); + if (kvm_enabled()) { + kvm_init_ap(); + if (kvm_irqchip && !kvm_has_gsi_routing(kvm_context)) { + irq0override = 0; + } + } machine->init(ram_size, vga_ram_size, boot_devices, kernel_filename, kernel_cmdline, initrd_filename, cpu_model);