Skip to contentSkip to conversation

x86/mm: drop 4MB restriction on minimal NUMA node size

PATCH Applied Mike Rapoport 5 participants 12 messages Oct 17, 2023 – Oct 20, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/mm: drop 4MB restriction on minimal NUMA node size

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

For example, if we use qemu to start a two NUMA node kernel, one of the nodes has 2M memory (less than NODE_MIN_SIZE), and the other node has 2G, then we will encounter the following panic:

  [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
  [    0.150783] #PF: supervisor write access in kernel mode
  [    0.151488] #PF: error_code(0x0002) - not-present page
  <...>
  [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
  <...>
  [    0.169781] Call Trace:
  [    0.170159]  <TASK>
  [    0.170448]  deactivate_slab+0x187/0x3c0
  [    0.171031]  ? bootstrap+0x1b/0x10e
  [    0.171559]  ? preempt_count_sub+0x9/0xa0
  [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
  [    0.172735]  ? bootstrap+0x1b/0x10e
  [    0.173236]  bootstrap+0x6b/0x10e
  [    0.173720]  kmem_cache_init+0x10a/0x188
  [    0.174240]  start_kernel+0x415/0x6ac
  [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
  [    0.175417]  </TASK>
  [    0.175713] Modules linked in:
  [    0.176117] CR2: 0000000000000000

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Drop the limitation for the minimal node size.

---
 arch/x86/include/asm/numa.h | 7 -------
 arch/x86/mm/numa.c          | 7 -------
 2 files changed, 14 deletions(-)
arch/x86/include/asm/numa.h +0 -7
index e3bae2b60a0d..ef2844d69173 100644
@@ -12,13 +12,6 @@
 
 #define NR_NODE_MEMBLKS		(MAX_NUMNODES*2)
 
-/*
- * Too small node sizes may confuse the VM badly. Usually they
- * result from BIOS bugs. So dont recognize nodes as standalone
- * NUMA entities that have less than this amount of RAM listed:
- */
-#define NODE_MIN_SIZE (4*1024*1024)
-
 extern int numa_off;
 
 /*
arch/x86/mm/numa.c +0 -7
index 2aadb2019b4f..55e3d895f15c 100644
@@ -601,13 +601,6 @@ static int __init numa_register_memblks(struct numa_meminfo *mi)
 		if (start >= end)
 			continue;
 
-		/*
-		 * Don't confuse VM with a node that doesn't have the
-		 * minimum amount of memory:
-		 */
-		if (end && (end - start) < NODE_MIN_SIZE)
-			continue;
-
 		alloc_node_data(nid);
 	}
 

base-commit: 94f6f0550c625fab1f373bb86a6669b45e9748b3

Signature · 1 line
2.39.2
#1 lore raw
Quoted 43 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

   For example, if we use qemu to start a two NUMA node kernel,
   one of the nodes has 2M memory (less than NODE_MIN_SIZE),
   and the other node has 2G, then we will encounter the
   following panic:

   [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
   [    0.150783] #PF: supervisor write access in kernel mode
   [    0.151488] #PF: error_code(0x0002) - not-present page
   <...>
   [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
   <...>
   [    0.169781] Call Trace:
   [    0.170159]  <TASK>
   [    0.170448]  deactivate_slab+0x187/0x3c0
   [    0.171031]  ? bootstrap+0x1b/0x10e
   [    0.171559]  ? preempt_count_sub+0x9/0xa0
   [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
   [    0.172735]  ? bootstrap+0x1b/0x10e
   [    0.173236]  bootstrap+0x6b/0x10e
   [    0.173720]  kmem_cache_init+0x10a/0x188
   [    0.174240]  start_kernel+0x415/0x6ac
   [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
   [    0.175417]  </TASK>
   [    0.175713] Modules linked in:
   [    0.176117] CR2: 0000000000000000

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Quoted 7 lines

Drop the limitation for the minimal node size.

That's just a resend I assume? Or has anything changed?

Signature · 3 lines
Cheers,

David / dhildenb
#2 lore raw
Quoted 46 lines · David Hildenbrand
Quoted 45 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

    For example, if we use qemu to start a two NUMA node kernel,
    one of the nodes has 2M memory (less than NODE_MIN_SIZE),
    and the other node has 2G, then we will encounter the
    following panic:

    [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
    [    0.150783] #PF: supervisor write access in kernel mode
    [    0.151488] #PF: error_code(0x0002) - not-present page
    <...>
    [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
    <...>
    [    0.169781] Call Trace:
    [    0.170159]  <TASK>
    [    0.170448]  deactivate_slab+0x187/0x3c0
    [    0.171031]  ? bootstrap+0x1b/0x10e
    [    0.171559]  ? preempt_count_sub+0x9/0xa0
    [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
    [    0.172735]  ? bootstrap+0x1b/0x10e
    [    0.173236]  bootstrap+0x6b/0x10e
    [    0.173720]  kmem_cache_init+0x10a/0x188
    [    0.174240]  start_kernel+0x415/0x6ac
    [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
    [    0.175417]  </TASK>
    [    0.175713] Modules linked in:
    [    0.176117] CR2: 0000000000000000

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Drop the limitation for the minimal node size.

Quoted 7 lines
Quoted 5 lines

That's just a resend I assume? Or has anything changed?

Saw the other mail now, so just a resend.

Signature · 3 lines
Cheers,

David / dhildenb
#3 lore raw
In reply to David Hildenbrand
Quoted 46 lines · David Hildenbrand
Quoted 45 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

   For example, if we use qemu to start a two NUMA node kernel,
   one of the nodes has 2M memory (less than NODE_MIN_SIZE),
   and the other node has 2G, then we will encounter the
   following panic:

   [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
   [    0.150783] #PF: supervisor write access in kernel mode
   [    0.151488] #PF: error_code(0x0002) - not-present page
   <...>
   [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
   <...>
   [    0.169781] Call Trace:
   [    0.170159]  <TASK>
   [    0.170448]  deactivate_slab+0x187/0x3c0
   [    0.171031]  ? bootstrap+0x1b/0x10e
   [    0.171559]  ? preempt_count_sub+0x9/0xa0
   [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
   [    0.172735]  ? bootstrap+0x1b/0x10e
   [    0.173236]  bootstrap+0x6b/0x10e
   [    0.173720]  kmem_cache_init+0x10a/0x188
   [    0.174240]  start_kernel+0x415/0x6ac
   [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
   [    0.175417]  </TASK>
   [    0.175713] Modules linked in:
   [    0.176117] CR2: 0000000000000000

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Drop the limitation for the minimal node size.

Quoted 7 lines
Quoted 5 lines

That's just a resend I assume? Or has anything changed?

Oh, I forgot RESEND prefix, sorry

Quoted 5 lines
Signature · 3 lines
Cheers,

David / dhildenb
Signature · 2 lines
Sincerely yours,
Mike.
#4 lore raw

PATCH v2 1/1 x86/mm: Drop 4MB restriction on minimal NUMA node memory size

In reply to Mike Rapoport
Quoted 31 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

For example, if we use qemu to start a two NUMA node kernel, one of the nodes has 2M memory (less than NODE_MIN_SIZE), and the other node has 2G, then we will encounter the following panic:

  [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
  [    0.150783] #PF: supervisor write access in kernel mode
  [    0.151488] #PF: error_code(0x0002) - not-present page
  <...>
  [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
  <...>
  [    0.169781] Call Trace:
  [    0.170159]  <TASK>
  [    0.170448]  deactivate_slab+0x187/0x3c0
  [    0.171031]  ? bootstrap+0x1b/0x10e
  [    0.171559]  ? preempt_count_sub+0x9/0xa0
  [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
  [    0.172735]  ? bootstrap+0x1b/0x10e
  [    0.173236]  bootstrap+0x6b/0x10e
  [    0.173720]  kmem_cache_init+0x10a/0x188
  [    0.174240]  start_kernel+0x415/0x6ac
  [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
  [    0.175417]  </TASK>
  [    0.175713] Modules linked in:
  [    0.176117] CR2: 0000000000000000
Quoted 3 lines

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

Presumably the core MM got fixed too to not just crash, but provide some sort of warning?

Quoted 7 lines

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Core MM won't get confused ... other than by the above weird Qemu topology, to which it responds with a ... NULL pointer dereference?

Seems quite close to the literal definition of 'get confused badly' to me, and doesn't give me the warm fuzzy feeling that giving the core MM even *more* weird topologies is super safe ... :-/

Quoted 1 line

Drop the limitation for the minimal node size.

While I agree with dropping the limitation, and I agree that 9391a3f9c7f1 should have provided more of a justification, I believe a core MM fix is in order as well, for it to not crash. [ If it's fixed upstream already, please reference the relevant commit ID. ]

Also, the changelog spelling & general presentation were quite low quality
- I've fixed it up a bit below, please carry this version going forward. Please spell-check your patches before sending out Nth versions of it, maybe maintainers are skipping them for a reason!

Thanks,

Ingo

=================>
From: "Mike Rapoport (IBM)" <[email protected]>
Date: Tue, 17 Oct 2023 09:22:15 +0300
Subject: [PATCH] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Qi Zheng reported crashes in a production environment and provided a simplified example as a reproducer:

 |  For example, if we use qemu to start a two NUMA node kernel,
 |  one of the nodes has 2M memory (less than NODE_MIN_SIZE),
 |  and the other node has 2G, then we will encounter the
 |  following panic:
 |
 |    BUG: kernel NULL pointer dereference, address: 0000000000000000
 |    <...>
 |    RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
 |    <...>
 |    Call Trace:
 |      <TASK>
 |      deactivate_slab()
 |      bootstrap()
 |      kmem_cache_init()
 |      start_kernel()
 |      secondary_startup_64_no_verify()

The crashes happen because of inconsistency between the nodemask that has nodes with less than 4MB as memoryless, and the actual memory fed into the core mm.

The commit:

9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing")

... that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

In the 17 years since then a lot has changed and core mm won't get confused about small node sizes.

Drop the limitation for the minimal node size.

[ mingo: Improved changelog clarity. ]

---
 arch/x86/include/asm/numa.h | 7 -------
 arch/x86/mm/numa.c          | 7 -------
 2 files changed, 14 deletions(-)
arch/x86/include/asm/numa.h +0 -7
index e3bae2b60a0d..ef2844d69173 100644
@@ -12,13 +12,6 @@
 
 #define NR_NODE_MEMBLKS		(MAX_NUMNODES*2)
 
-/*
- * Too small node sizes may confuse the VM badly. Usually they
- * result from BIOS bugs. So dont recognize nodes as standalone
- * NUMA entities that have less than this amount of RAM listed:
- */
-#define NODE_MIN_SIZE (4*1024*1024)
-
 extern int numa_off;
 
 /*
arch/x86/mm/numa.c +0 -7
index c01c5506fd4a..aa39d678fe81 100644
@@ -602,13 +602,6 @@ static int __init numa_register_memblks(struct numa_meminfo *mi)
 		if (start >= end)
 			continue;
 
-		/*
-		 * Don't confuse VM with a node that doesn't have the
-		 * minimum amount of memory:
-		 */
-		if (end && (end - start) < NODE_MIN_SIZE)
-			continue;
-
 		alloc_node_data(nid);
 	}
 
#5 lore raw

Re: [PATCH v2] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Hi all,

Quoted 56 lines · Ingo Molnar
Quoted 31 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

   For example, if we use qemu to start a two NUMA node kernel,
   one of the nodes has 2M memory (less than NODE_MIN_SIZE),
   and the other node has 2G, then we will encounter the
   following panic:

   [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
   [    0.150783] #PF: supervisor write access in kernel mode
   [    0.151488] #PF: error_code(0x0002) - not-present page
   <...>
   [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
   <...>
   [    0.169781] Call Trace:
   [    0.170159]  <TASK>
   [    0.170448]  deactivate_slab+0x187/0x3c0
   [    0.171031]  ? bootstrap+0x1b/0x10e
   [    0.171559]  ? preempt_count_sub+0x9/0xa0
   [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
   [    0.172735]  ? bootstrap+0x1b/0x10e
   [    0.173236]  bootstrap+0x6b/0x10e
   [    0.173720]  kmem_cache_init+0x10a/0x188
   [    0.174240]  start_kernel+0x415/0x6ac
   [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
   [    0.175417]  </TASK>
   [    0.175713] Modules linked in:
   [    0.176117] CR2: 0000000000000000
Quoted 3 lines

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

Presumably the core MM got fixed too to not just crash, but provide some sort of warning?

Quoted 7 lines

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Core MM won't get confused ... other than by the above weird Qemu topology, to which it responds with a ... NULL pointer dereference?

Seems quite close to the literal definition of 'get confused badly' to me, and doesn't give me the warm fuzzy feeling that giving the core MM even *more* weird topologies is super safe ... :-/

Quoted 6 lines
Quoted 1 line

Drop the limitation for the minimal node size.

While I agree with dropping the limitation, and I agree that 9391a3f9c7f1 should have provided more of a justification, I believe a core MM fix is in order as well, for it to not crash. [ If it's fixed upstream already, please reference the relevant commit ID. ]

Agree. I posted a fixed patchset[1] before, maybe we can reconsider it. :)

[1]. https://lore.kernel.org/lkml/[email protected]/

For memoryless node, this patchset skip it and fallback to other nodes when build its zonelists.

Say we have node0 and node1, and node0 is memoryless, then:

[    0.102400] Fallback order for Node 0: 1
[    0.102931] Fallback order for Node 1: 1

In this way, we will not allocate pages from memoryless node0. Then the crash problem under the weird Qemu topology will be fixed.

Thanks,
Qi

Quoted 101 lines

Also, the changelog spelling & general presentation were quite low quality
- I've fixed it up a bit below, please carry this version going forward. Please spell-check your patches before sending out Nth versions of it, maybe maintainers are skipping them for a reason!

Thanks,

Ingo

=================>
From: "Mike Rapoport (IBM)" <[email protected]>
Date: Tue, 17 Oct 2023 09:22:15 +0300
Subject: [PATCH] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Qi Zheng reported crashes in a production environment and provided a simplified example as a reproducer:

  |  For example, if we use qemu to start a two NUMA node kernel,
  |  one of the nodes has 2M memory (less than NODE_MIN_SIZE),
  |  and the other node has 2G, then we will encounter the
  |  following panic:
  |
  |    BUG: kernel NULL pointer dereference, address: 0000000000000000
  |    <...>
  |    RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
  |    <...>
  |    Call Trace:
  |      <TASK>
  |      deactivate_slab()
  |      bootstrap()
  |      kmem_cache_init()
  |      start_kernel()
  |      secondary_startup_64_no_verify()

The crashes happen because of inconsistency between the nodemask that has nodes with less than 4MB as memoryless, and the actual memory fed into the core mm.

The commit:

9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing")

... that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

In the 17 years since then a lot has changed and core mm won't get confused about small node sizes.

Drop the limitation for the minimal node size.

[ mingo: Improved changelog clarity. ]

---
  arch/x86/include/asm/numa.h | 7 -------
  arch/x86/mm/numa.c          | 7 -------
  2 files changed, 14 deletions(-)
arch/x86/include/asm/numa.h +0 -7
index e3bae2b60a0d..ef2844d69173 100644
@@ -12,13 +12,6 @@
 
  #define NR_NODE_MEMBLKS		(MAX_NUMNODES*2)
 
-/*
- * Too small node sizes may confuse the VM badly. Usually they
- * result from BIOS bugs. So dont recognize nodes as standalone
- * NUMA entities that have less than this amount of RAM listed:
- */
-#define NODE_MIN_SIZE (4*1024*1024)
-
  extern int numa_off;
 
  /*
arch/x86/mm/numa.c +0 -7
index c01c5506fd4a..aa39d678fe81 100644
@@ -602,13 +602,6 @@ static int __init numa_register_memblks(struct numa_meminfo *mi)
  		if (start >= end)
  			continue;
 
-		/*
-		 * Don't confuse VM with a node that doesn't have the
-		 * minimum amount of memory:
-		 */
-		if (end && (end - start) < NODE_MIN_SIZE)
-			continue;
-
  		alloc_node_data(nid);
  	}
 
#6 lore raw

Re: [PATCH v2] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Quoted 5 lines · Qi Zheng
Quoted 4 lines

While I agree with dropping the limitation, and I agree that 9391a3f9c7f1 should have provided more of a justification, I believe a core MM fix is in order as well, for it to not crash. [ If it's fixed upstream already, please reference the relevant commit ID. ]

Quoted 7 lines

Agree. I posted a fixed patchset[1] before, maybe we can reconsider it. :)

[1]. https://lore.kernel.org/lkml/[email protected]/

For memoryless node, this patchset skip it and fallback to other nodes when build its zonelists.

Mind resubmitting that to the MM folks, with the NULL dereference crash mentioned prominently? Feel free to Cc: me.

Fixing hypothetical robustness problems is good, fixing specific crashes is better. :-)

Thanks,

Ingo

#7 lore raw

Re: [PATCH v2] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Hi Ingo,

Quoted 11 lines · Ingo Molnar
Quoted 8 lines · Qi Zheng
Quoted 4 lines

While I agree with dropping the limitation, and I agree that 9391a3f9c7f1 should have provided more of a justification, I believe a core MM fix is in order as well, for it to not crash. [ If it's fixed upstream already, please reference the relevant commit ID. ]

Agree. I posted a fixed patchset[1] before, maybe we can reconsider it. :)

Quoted 7 lines
Quoted 4 lines

[1]. https://lore.kernel.org/lkml/[email protected]/

For memoryless node, this patchset skip it and fallback to other nodes when build its zonelists.

Mind resubmitting that to the MM folks, with the NULL dereference crash mentioned prominently? Feel free to Cc: me.

OK, I will resend it if no one else objects. :)

Thanks,
Qi

Quoted 7 lines

Fixing hypothetical robustness problems is good, fixing specific crashes is better. :-)

Thanks,

Ingo

#8 lore raw

Re: [PATCH v2] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Quoted 16 lines · Qi Zheng

Hi Ingo,

Quoted 13 lines · Ingo Molnar
Quoted 10 lines · Qi Zheng
Quoted 4 lines

While I agree with dropping the limitation, and I agree that 9391a3f9c7f1 should have provided more of a justification, I believe a core MM fix is in order as well, for it to not crash. [ If it's fixed upstream already, please reference the relevant commit ID. ]

Agree. I posted a fixed patchset[1] before, maybe we can reconsider it. :)

[1]. https://lore.kernel.org/lkml/[email protected]/

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

For memoryless node, this patchset skip it and fallback to other nodes when build its zonelists.

Mind resubmitting that to the MM folks, with the NULL dereference crash mentioned prominently? Feel free to Cc: me.

OK, I will resend it if no one else objects. :)

Thanks, much appreciated - and I see Andrew already applied your two fixes to -mm.

With that background I was able to apply the x86 fix as well - which can be backported without the MM changes. The current commit in tip:x86/mm is:

a1e2b8b36820 ("x86/mm: Drop the 4 MB restriction on minimal NUMA node memory size")

Which should hit v6.7 in about ~1.5 weeks, unless there's unexpected problems.

Thanks,

Ingo

#9 lore raw

Re: [PATCH v2] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Quoted 8 lines · Ingo Molnar
Quoted 4 lines
Quoted 2 lines

Mind resubmitting that to the MM folks, with the NULL dereference crash mentioned prominently? Feel free to Cc: me.

OK, I will resend it if no one else objects. :)

Thanks, much appreciated - and I see Andrew already applied your two fixes to -mm.

Quoted 8 lines

With that background I was able to apply the x86 fix as well - which can be backported without the MM changes. The current commit in tip:x86/mm is:

a1e2b8b36820 ("x86/mm: Drop the 4 MB restriction on minimal NUMA node memory size")

Which should hit v6.7 in about ~1.5 weeks, unless there's unexpected problems.

Note that I haven't added a Cc: stable tag, a 17 years old bug is not really a regression - but I have no objections against this fix getting into -stable once it gains a bit more testing and hits upstream.

Thanks,

Ingo

#10 lore raw

Re: [PATCH v2] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

In reply to Ingo Molnar
Quoted 49 lines · Ingo Molnar
Quoted 31 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

For example, if we use qemu to start a two NUMA node kernel, one of the nodes has 2M memory (less than NODE_MIN_SIZE), and the other node has 2G, then we will encounter the following panic:

  [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
  [    0.150783] #PF: supervisor write access in kernel mode
  [    0.151488] #PF: error_code(0x0002) - not-present page
  <...>
  [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
  <...>
  [    0.169781] Call Trace:
  [    0.170159]  <TASK>
  [    0.170448]  deactivate_slab+0x187/0x3c0
  [    0.171031]  ? bootstrap+0x1b/0x10e
  [    0.171559]  ? preempt_count_sub+0x9/0xa0
  [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
  [    0.172735]  ? bootstrap+0x1b/0x10e
  [    0.173236]  bootstrap+0x6b/0x10e
  [    0.173720]  kmem_cache_init+0x10a/0x188
  [    0.174240]  start_kernel+0x415/0x6ac
  [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
  [    0.175417]  </TASK>
  [    0.175713] Modules linked in:
  [    0.176117] CR2: 0000000000000000
Quoted 3 lines

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

Presumably the core MM got fixed too to not just crash, but provide some sort of warning?

Quoted 7 lines

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Quoted 6 lines

Core MM won't get confused ... other than by the above weird Qemu topology, to which it responds with a ... NULL pointer dereference?

Seems quite close to the literal definition of 'get confused badly' to me, and doesn't give me the warm fuzzy feeling that giving the core MM even *more* weird topologies is super safe ... :-/

The confusion is not about topology and not because of the small size of a node.
The confusion is because x86 fails to consistently report it's memory layout. At one point it says there is memoryless node and another point it says that that node actually has memory.

And dropping the kludge that says "Don't confuse VM with a node that doesn't have the minimum amount of memory" fixes exactly that.

Quoted 6 lines
Quoted 1 line

Drop the limitation for the minimal node size.

While I agree with dropping the limitation, and I agree that 9391a3f9c7f1 should have provided more of a justification, I believe a core MM fix is in order as well, for it to not crash. [ If it's fixed upstream already, please reference the relevant commit ID. ]

The core mm can be more strict about memory layouts it accepts indeed. I'll look into that.

Quoted 100 lines

Also, the changelog spelling & general presentation were quite low quality
- I've fixed it up a bit below, please carry this version going forward. Please spell-check your patches before sending out Nth versions of it, maybe maintainers are skipping them for a reason!

Thanks,

Ingo

=================>
From: "Mike Rapoport (IBM)" <[email protected]>
Date: Tue, 17 Oct 2023 09:22:15 +0300
Subject: [PATCH] x86/mm: Drop 4MB restriction on minimal NUMA node memory size

Qi Zheng reported crashes in a production environment and provided a simplified example as a reproducer:

 |  For example, if we use qemu to start a two NUMA node kernel,
 |  one of the nodes has 2M memory (less than NODE_MIN_SIZE),
 |  and the other node has 2G, then we will encounter the
 |  following panic:
 |
 |    BUG: kernel NULL pointer dereference, address: 0000000000000000
 |    <...>
 |    RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
 |    <...>
 |    Call Trace:
 |      <TASK>
 |      deactivate_slab()
 |      bootstrap()
 |      kmem_cache_init()
 |      start_kernel()
 |      secondary_startup_64_no_verify()

The crashes happen because of inconsistency between the nodemask that has nodes with less than 4MB as memoryless, and the actual memory fed into the core mm.

The commit:

9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing")

... that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

In the 17 years since then a lot has changed and core mm won't get confused about small node sizes.

Drop the limitation for the minimal node size.

[ mingo: Improved changelog clarity. ]

---
 arch/x86/include/asm/numa.h | 7 -------
 arch/x86/mm/numa.c          | 7 -------
 2 files changed, 14 deletions(-)
arch/x86/include/asm/numa.h +0 -7
index e3bae2b60a0d..ef2844d69173 100644
@@ -12,13 +12,6 @@
 
 #define NR_NODE_MEMBLKS		(MAX_NUMNODES*2)
 
-/*
- * Too small node sizes may confuse the VM badly. Usually they
- * result from BIOS bugs. So dont recognize nodes as standalone
- * NUMA entities that have less than this amount of RAM listed:
- */
-#define NODE_MIN_SIZE (4*1024*1024)
-
 extern int numa_off;
 
 /*
arch/x86/mm/numa.c +0 -7
index c01c5506fd4a..aa39d678fe81 100644
@@ -602,13 +602,6 @@ static int __init numa_register_memblks(struct numa_meminfo *mi)
 		if (start >= end)
 			continue;
 
-		/*
-		 * Don't confuse VM with a node that doesn't have the
-		 * minimum amount of memory:
-		 */
-		if (end && (end - start) < NODE_MIN_SIZE)
-			continue;
-
 		alloc_node_data(nid);
 	}
 
Signature · 2 lines
Sincerely yours,
Mike.
Tested-by on v1 1/1
#11 lore raw
In reply to Mike Rapoport

This patch has been successfully tested by QE. Start a VM with two NUMA nodes, one of them with less than 2M of memory. Check there is no kernel panic and the VM boots up smoothly.

BR,
Mario

Quoted 97 lines · Mike Rapoport

From: "Mike Rapoport (IBM)" <[email protected]>

Qi Zheng reports crashes in a production environment and provides a simplified example as a reproducer:

For example, if we use qemu to start a two NUMA node kernel, one of the nodes has 2M memory (less than NODE_MIN_SIZE), and the other node has 2G, then we will encounter the following panic:

  [    0.149844] BUG: kernel NULL pointer dereference, address: 0000000000000000
  [    0.150783] #PF: supervisor write access in kernel mode
  [    0.151488] #PF: error_code(0x0002) - not-present page
  <...>
  [    0.156056] RIP: 0010:_raw_spin_lock_irqsave+0x22/0x40
  <...>
  [    0.169781] Call Trace:
  [    0.170159]  <TASK>
  [    0.170448]  deactivate_slab+0x187/0x3c0
  [    0.171031]  ? bootstrap+0x1b/0x10e
  [    0.171559]  ? preempt_count_sub+0x9/0xa0
  [    0.172145]  ? kmem_cache_alloc+0x12c/0x440
  [    0.172735]  ? bootstrap+0x1b/0x10e
  [    0.173236]  bootstrap+0x6b/0x10e
  [    0.173720]  kmem_cache_init+0x10a/0x188
  [    0.174240]  start_kernel+0x415/0x6ac
  [    0.174738]  secondary_startup_64_no_verify+0xe0/0xeb
  [    0.175417]  </TASK>
  [    0.175713] Modules linked in:
  [    0.176117] CR2: 0000000000000000

The crashes happen because of inconsistency between nodemask that has nodes with less than 4MB as memoryless and the actual memory fed into core mm.

The commit 9391a3f9c7f1 ("[PATCH] x86_64: Clear more state when ignoring empty node in SRAT parsing") that introduced minimal size of a NUMA node does not explain why a node size cannot be less than 4MB and what boot failures this restriction might fix.

Since then a lot has changed and core mm won't confuse badly about small node sizes.

Drop the limitation for the minimal node size.

---
 arch/x86/include/asm/numa.h | 7 -------
 arch/x86/mm/numa.c          | 7 -------
 2 files changed, 14 deletions(-)
arch/x86/include/asm/numa.h +0 -7
index e3bae2b60a0d..ef2844d69173 100644
@@ -12,13 +12,6 @@
 
 #define NR_NODE_MEMBLKS                (MAX_NUMNODES*2)
 
-/*
- * Too small node sizes may confuse the VM badly. Usually they
- * result from BIOS bugs. So dont recognize nodes as standalone
- * NUMA entities that have less than this amount of RAM listed:
- */
-#define NODE_MIN_SIZE (4*1024*1024)
-
 extern int numa_off;
 
 /*
arch/x86/mm/numa.c +0 -7
index 2aadb2019b4f..55e3d895f15c 100644
@@ -601,13 +601,6 @@ static int __init numa_register_memblks(struct numa_meminfo *mi)
                if (start >= end)
                        continue;
 
-               /*
-                * Don't confuse VM with a node that doesn't have the
-                * minimum amount of memory:
-                */
-               if (end && (end - start) < NODE_MIN_SIZE)
-                       continue;
-
                alloc_node_data(nid);
        }
 

base-commit: 94f6f0550c625fab1f373bb86a6669b45e9748b3 --
2.39.2