diff mbox series

[v2,4/5] mm/page_alloc: place pages to tail in __free_pages_core()

Message ID 20201005121534.15649-5-david@redhat.com (mailing list archive)
State Not Applicable, archived
Headers show
Series mm: place pages to the freelist tail when onlining and undoing isolation | expand

Commit Message

David Hildenbrand Oct. 5, 2020, 12:15 p.m. UTC 
__free_pages_core() is used when exposing fresh memory to the buddy
during system boot and when onlining memory in generic_online_page().

generic_online_page() is used in two cases:

1. Direct memory onlining in online_pages().
2. Deferred memory onlining in memory-ballooning-like mechanisms (HyperV
   balloon and virtio-mem), when parts of a section are kept
   fake-offline to be fake-onlined later on.

In 1, we already place pages to the tail of the freelist. Pages will be
freed to MIGRATE_ISOLATE lists first and moved to the tail of the freelists
via undo_isolate_page_range().

In 2, we currently don't implement a proper rule. In case of virtio-mem,
where we currently always online MAX_ORDER - 1 pages, the pages will be
placed to the HEAD of the freelist - undesireable. While the hyper-v
balloon calls generic_online_page() with single pages, usually it will
call it on successive single pages in a larger block.

The pages are fresh, so place them to the tail of the freelist and avoid
the PCP. In __free_pages_core(), remove the now superflouos call to
set_page_refcounted() and add a comment regarding page initialization and
the refcount.

Note: In 2. we currently don't shuffle. If ever relevant (page shuffling
is usually of limited use in virtualized environments), we might want to
shuffle after a sequence of generic_online_page() calls in the
relevant callers.

Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Reviewed-by: Wei Yang <richard.weiyang@linux.alibaba.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Wei Liu <wei.liu@kernel.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 mm/page_alloc.c | 33 +++++++++++++++++++++++----------
 1 file changed, 23 insertions(+), 10 deletions(-)

Comments

David Hildenbrand Sept. 8, 2021, 6:21 a.m. UTC | #1 
On 08.09.21 00:40, Sean Anderson wrote:
> Hi David,
> 
> This patch breaks booting on my custom Xilinx ZynqMP board. Booting
> fails just after/during GIC initialization:
> 
> [    0.000000] Booting Linux on physical CPU 0x0000000000 [0x410fd034]
> [    0.000000] Linux version 5.14.0 (sean@plantagenet) (aarch64-linux-gnu-gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0, GNU ld (GNU Binutils for Ubuntu) 2.34) #251 SMP Tue Sep 7 18:11:50 EDT 2021
> [    0.000000] Machine model: xlnx,zynqmp
> [    0.000000] earlycon: cdns0 at MMIO 0x00000000ff010000 (options '115200n8')
> [    0.000000] printk: bootconsole [cdns0] enabled
> [    0.000000] efi: UEFI not found.
> [    0.000000] Zone ranges:
> [    0.000000]   DMA32    [mem 0x0000000000000000-0x00000000ffffffff]
> [    0.000000]   Normal   [mem 0x0000000100000000-0x000000087fffffff]
> [    0.000000] Movable zone start for each node
> [    0.000000] Early memory node ranges
> [    0.000000]   node   0: [mem 0x0000000000000000-0x000000007fefffff]
> [    0.000000]   node   0: [mem 0x0000000800000000-0x000000087fffffff]
> [    0.000000] Initmem setup node 0 [mem 0x0000000000000000-0x000000087fffffff]
> [    0.000000] On node 0, zone Normal: 256 pages in unavailable ranges
> [    0.000000] cma: Reserved 1000 MiB at 0x0000000041400000
> [    0.000000] psci: probing for conduit method from DT.
> [    0.000000] psci: PSCIv1.1 detected in firmware.
> [    0.000000] psci: Using standard PSCI v0.2 function IDs
> [    0.000000] psci: MIGRATE_INFO_TYPE not supported.
> [    0.000000] psci: SMC Calling Convention v1.1
> [    0.000000] percpu: Embedded 19 pages/cpu s46752 r0 d31072 u77824
> [    0.000000] Detected VIPT I-cache on CPU0
> [    0.000000] CPU features: detected: ARM erratum 845719
> [    0.000000] Built 1 zonelists, mobility grouping on.  Total pages: 1033987
> [    0.000000] Kernel command line: earlycon clk_ignore_unused root=/dev/mmcblk0p2 rootwait rw cma=1000M
> [    0.000000] Dentry cache hash table entries: 524288 (order: 10, 4194304 bytes, linear)
> [    0.000000] Inode-cache hash table entries: 262144 (order: 9, 2097152 bytes, linear)
> [    0.000000] mem auto-init: stack:off, heap alloc:off, heap free:off
> [    0.000000] software IO TLB: mapped [mem 0x000000003d400000-0x0000000041400000] (64MB)
> [    0.000000] Memory: 3023384K/4193280K available (4288K kernel code, 514K rwdata, 1200K rodata, 896K init, 187K bss, 145896K reserved, 1024000K cma-reserved)
> [    0.000000] rcu: Hierarchical RCU implementation.
> [    0.000000] rcu: 	RCU event tracing is enabled.
> [    0.000000] rcu: RCU calculated value of scheduler-enlistment delay is 25 jiffies.
> [    0.000000] NR_IRQS: 64, nr_irqs: 64, preallocated irqs: 0
> [    0.000000] GIC: Adjusting CPU interface base to 0x00000000f902f000
> [    0.000000] Root IRQ handler: gic_handle_irq
> [    0.000000] GIC: Using split EOI/Deactivate mode
> 
> and I bisected it to this patch. Applying the following patch (for 5.14)
> fixes booting again:

Hi Sean,

unfortunately that patch most likely (with 99.9999% confidence) revealed 
another latent BUG in your setup.

Some memory that shouldn't be handed to the buddy as free memory is 
getting now allocated earlier than later, resulting in that issue.


I had all different kinds of reports, but they were mostly

a) Firmware bugs that result in uncached memory getting exposed to the 
buddy, resulting in severe performance degradation such that the system 
will no longer boot. [3]

I wrote kstream [1] to be run under the old kernel, to identify these.

b) BUGs that result in unsuitable memory getting exposed to either the 
buddy or devices, resulting in errors during device initialization. [6]

c) Use after free BUGs.

Exposing memory, such as used for ACPI tables, to the buddy as free 
memory although it's still in use. [4]

d) Hypervisor BUGs

The last report (heavy performance degradation) was due to a BUG in 
dpdk. [2]


What the exact symptoms you're experiencing? Really slow boot/stall? 
Then it could be a) and kstream might help.


[1] https://github.com/davidhildenbrand/kstream
[2] 
https://lore.kernel.org/dpdk-dev/20210827161231.579968-1-eperezma@redhat.com/T/#u
[3] 
https://lore.kernel.org/r/MW3PR12MB4537C3C6EFD9CA3A4B32084DF36B9@MW3PR12MB4537.namprd12.prod.outlook.com
[4] https://lkml.kernel.org/r/4650320.31r3eYUQgx@kreacher
[5] https://lkml.kernel.org/r/87361onphy.fsf_-_@codeaurora.org
[6] 
https://lore.kernel.org/r/20201213225517.3838501-1-linus.walleij@linaro.org
Sean Anderson Sept. 10, 2021, 11:06 p.m. UTC | #2 
On 9/8/21 2:21 AM, David Hildenbrand wrote:
> On 08.09.21 00:40, Sean Anderson wrote:
>> Hi David,
>>
>> This patch breaks booting on my custom Xilinx ZynqMP board. Booting
>> fails just after/during GIC initialization:
>>
>> [    0.000000] Booting Linux on physical CPU 0x0000000000 [0x410fd034]
>> [    0.000000] Linux version 5.14.0 (sean@plantagenet) (aarch64-linux-gnu-gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0, GNU ld (GNU Binutils for Ubuntu) 2.34) #251 SMP Tue Sep 7 18:11:50 EDT 2021
>> [    0.000000] Machine model: xlnx,zynqmp
>> [    0.000000] earlycon: cdns0 at MMIO 0x00000000ff010000 (options '115200n8')
>> [    0.000000] printk: bootconsole [cdns0] enabled
>> [    0.000000] efi: UEFI not found.
>> [    0.000000] Zone ranges:
>> [    0.000000]   DMA32    [mem 0x0000000000000000-0x00000000ffffffff]
>> [    0.000000]   Normal   [mem 0x0000000100000000-0x000000087fffffff]
>> [    0.000000] Movable zone start for each node
>> [    0.000000] Early memory node ranges
>> [    0.000000]   node   0: [mem 0x0000000000000000-0x000000007fefffff]
>> [    0.000000]   node   0: [mem 0x0000000800000000-0x000000087fffffff]
>> [    0.000000] Initmem setup node 0 [mem 0x0000000000000000-0x000000087fffffff]
>> [    0.000000] On node 0, zone Normal: 256 pages in unavailable ranges
>> [    0.000000] cma: Reserved 1000 MiB at 0x0000000041400000
>> [    0.000000] psci: probing for conduit method from DT.
>> [    0.000000] psci: PSCIv1.1 detected in firmware.
>> [    0.000000] psci: Using standard PSCI v0.2 function IDs
>> [    0.000000] psci: MIGRATE_INFO_TYPE not supported.
>> [    0.000000] psci: SMC Calling Convention v1.1
>> [    0.000000] percpu: Embedded 19 pages/cpu s46752 r0 d31072 u77824
>> [    0.000000] Detected VIPT I-cache on CPU0
>> [    0.000000] CPU features: detected: ARM erratum 845719
>> [    0.000000] Built 1 zonelists, mobility grouping on.  Total pages: 1033987
>> [    0.000000] Kernel command line: earlycon clk_ignore_unused root=/dev/mmcblk0p2 rootwait rw cma=1000M
>> [    0.000000] Dentry cache hash table entries: 524288 (order: 10, 4194304 bytes, linear)
>> [    0.000000] Inode-cache hash table entries: 262144 (order: 9, 2097152 bytes, linear)
>> [    0.000000] mem auto-init: stack:off, heap alloc:off, heap free:off
>> [    0.000000] software IO TLB: mapped [mem 0x000000003d400000-0x0000000041400000] (64MB)
>> [    0.000000] Memory: 3023384K/4193280K available (4288K kernel code, 514K rwdata, 1200K rodata, 896K init, 187K bss, 145896K reserved, 1024000K cma-reserved)
>> [    0.000000] rcu: Hierarchical RCU implementation.
>> [    0.000000] rcu:     RCU event tracing is enabled.
>> [    0.000000] rcu: RCU calculated value of scheduler-enlistment delay is 25 jiffies.
>> [    0.000000] NR_IRQS: 64, nr_irqs: 64, preallocated irqs: 0
>> [    0.000000] GIC: Adjusting CPU interface base to 0x00000000f902f000
>> [    0.000000] Root IRQ handler: gic_handle_irq
>> [    0.000000] GIC: Using split EOI/Deactivate mode
>>
>> and I bisected it to this patch. Applying the following patch (for 5.14)
>> fixes booting again:
>
> Hi Sean,
>
> unfortunately that patch most likely (with 99.9999% confidence) revealed another latent BUG in your setup.

I suspected as much; however even after inspecting this patch I was
unsure what I should investigate further.

>
> Some memory that shouldn't be handed to the buddy as free memory is getting now allocated earlier than later, resulting in that issue.
>
>
> I had all different kinds of reports, but they were mostly
>
> a) Firmware bugs that result in uncached memory getting exposed to the buddy, resulting in severe performance degradation such that the system will no longer boot. [3]
>
> I wrote kstream [1] to be run under the old kernel, to identify these.
>
> b) BUGs that result in unsuitable memory getting exposed to either the buddy or devices, resulting in errors during device initialization. [6]
>
> c) Use after free BUGs.
>
> Exposing memory, such as used for ACPI tables, to the buddy as free memory although it's still in use. [4]
>
> d) Hypervisor BUGs
>
> The last report (heavy performance degradation) was due to a BUG in dpdk. [2]
>
>
> What the exact symptoms you're experiencing? Really slow boot/stall? Then it could be a) and kstream might help.

Well, as it turns out DDR chips of half the correct size were installed.
This caused the upper half of memory to alias to the lower half. As it
happened, due to some lucky circumstances this didn't initially cause
problems. Sorry for the noise.

--Sean
diff mbox series

Patch

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index b187e46cf640..3dadcc6d4009 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -275,7 +275,8 @@  bool pm_suspended_storage(void)
 unsigned int pageblock_order __read_mostly;
 #endif
 
-static void __free_pages_ok(struct page *page, unsigned int order);
+static void __free_pages_ok(struct page *page, unsigned int order,
+			    fpi_t fpi_flags);
 
 /*
  * results with 256, 32 in the lowmem_reserve sysctl:
@@ -687,7 +688,7 @@  static void bad_page(struct page *page, const char *reason)
 void free_compound_page(struct page *page)
 {
 	mem_cgroup_uncharge(page);
-	__free_pages_ok(page, compound_order(page));
+	__free_pages_ok(page, compound_order(page), FPI_NONE);
 }
 
 void prep_compound_page(struct page *page, unsigned int order)
@@ -1423,14 +1424,14 @@  static void free_pcppages_bulk(struct zone *zone, int count,
 static void free_one_page(struct zone *zone,
 				struct page *page, unsigned long pfn,
 				unsigned int order,
-				int migratetype)
+				int migratetype, fpi_t fpi_flags)
 {
 	spin_lock(&zone->lock);
 	if (unlikely(has_isolate_pageblock(zone) ||
 		is_migrate_isolate(migratetype))) {
 		migratetype = get_pfnblock_migratetype(page, pfn);
 	}
-	__free_one_page(page, pfn, zone, order, migratetype, FPI_NONE);
+	__free_one_page(page, pfn, zone, order, migratetype, fpi_flags);
 	spin_unlock(&zone->lock);
 }
 
@@ -1508,7 +1509,8 @@  void __meminit reserve_bootmem_region(phys_addr_t start, phys_addr_t end)
 	}
 }
 
-static void __free_pages_ok(struct page *page, unsigned int order)
+static void __free_pages_ok(struct page *page, unsigned int order,
+			    fpi_t fpi_flags)
 {
 	unsigned long flags;
 	int migratetype;
@@ -1520,7 +1522,8 @@  static void __free_pages_ok(struct page *page, unsigned int order)
 	migratetype = get_pfnblock_migratetype(page, pfn);
 	local_irq_save(flags);
 	__count_vm_events(PGFREE, 1 << order);
-	free_one_page(page_zone(page), page, pfn, order, migratetype);
+	free_one_page(page_zone(page), page, pfn, order, migratetype,
+		      fpi_flags);
 	local_irq_restore(flags);
 }
 
@@ -1530,6 +1533,11 @@  void __free_pages_core(struct page *page, unsigned int order)
 	struct page *p = page;
 	unsigned int loop;
 
+	/*
+	 * When initializing the memmap, __init_single_page() sets the refcount
+	 * of all pages to 1 ("allocated"/"not free"). We have to set the
+	 * refcount of all involved pages to 0.
+	 */
 	prefetchw(p);
 	for (loop = 0; loop < (nr_pages - 1); loop++, p++) {
 		prefetchw(p + 1);
@@ -1540,8 +1548,12 @@  void __free_pages_core(struct page *page, unsigned int order)
 	set_page_count(p, 0);
 
 	atomic_long_add(nr_pages, &page_zone(page)->managed_pages);
-	set_page_refcounted(page);
-	__free_pages(page, order);
+
+	/*
+	 * Bypass PCP and place fresh pages right to the tail, primarily
+	 * relevant for memory onlining.
+	 */
+	__free_pages_ok(page, order, FPI_TO_TAIL);
 }
 
 #ifdef CONFIG_NEED_MULTIPLE_NODES
@@ -3168,7 +3180,8 @@  static void free_unref_page_commit(struct page *page, unsigned long pfn)
 	 */
 	if (migratetype >= MIGRATE_PCPTYPES) {
 		if (unlikely(is_migrate_isolate(migratetype))) {
-			free_one_page(zone, page, pfn, 0, migratetype);
+			free_one_page(zone, page, pfn, 0, migratetype,
+				      FPI_NONE);
 			return;
 		}
 		migratetype = MIGRATE_MOVABLE;
@@ -4991,7 +5004,7 @@  static inline void free_the_page(struct page *page, unsigned int order)
 	if (order == 0)		/* Via pcp? */
 		free_unref_page(page);
 	else
-		__free_pages_ok(page, order);
+		__free_pages_ok(page, order, FPI_NONE);
 }
 
 void __free_pages(struct page *page, unsigned int order)