Skip to contentSkip to conversation

mm/vmalloc: clarify why vmap_range_noflush() might sleep

PATCH Superseded Brendan Jackman 5 participants 8 messages Dec 8, 2025 – Dec 9, 2025

Subsystems (2)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mm/vmalloc: clarify why vmap_range_noflush() might sleep

The only reason vmap_range_noflush() can sleep is because of pagetable allocations. This might_sleep() is accurate, but we have a more precise way to express this particular case, so help readers out by using that.

Note that the actual GFP flags used to allocate here are arch-specific. But as long as GFP_PGTABLE_KERNEL includes blockable flags, it should serve as a reasonable common-denominator here.

This also ensures there is an fs_reclaim_acquire() even no pagetables are actually allocated, which could potentially do a better job at catching filesystem bugs.

---

---
 mm/vmalloc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/vmalloc.c +1 -1
index a53c7462671bdd896f95712af71398ffbe22fb80..ff1876588b94ec69168324e93399dbd117a6959a 100644
@@ -305,7 +305,7 @@ static int vmap_range_noflush(unsigned long addr, unsigned long end,
 	int err;
 	pgtbl_mod_mask mask = 0;
 
-	might_sleep();
+	might_alloc(GFP_PGTABLE_KERNEL);
 	BUG_ON(addr >= end);
 
 	start = addr;

---
base-commit: ecc46e02e0abe025a6e840cba2d647f23fd1d721
change-id: 20251208-b4-vmalloc-might_alloc-754a791e4e10

Best regards,

Signature · 1 line
Brendan Jackman <[email protected]>
Reviewed-by on v1 1/1
#1 lore raw
Quoted 7 lines · Brendan Jackman

The only reason vmap_range_noflush() can sleep is because of pagetable allocations. This might_sleep() is accurate, but we have a more precise way to express this particular case, so help readers out by using that.

Note that the actual GFP flags used to allocate here are arch-specific. But as long as GFP_PGTABLE_KERNEL includes blockable flags, it should

Currently GFP_PGTABLE_KERNEL does have a blocking flag via GFP_KERNEL.

#define GFP_PGTABLE_KERNEL (GFP_KERNEL | __GFP_ZERO)
#define GFP_KERNEL (__GFP_RECLAIM | __GFP_IO | __GFP_FS)
#define __GFP_RECLAIM ((__force gfp_t)(___GFP_DIRECT_RECLAIM|___GFP_KSWAPD_RECLAIM))

Quoted 1 line

serve as a reasonable common-denominator here.

Agreed.

Quoted 4 lines

This also ensures there is an fs_reclaim_acquire() even no pagetables are actually allocated, which could potentially do a better job at catching filesystem bugs.

Makes sense.

Quoted 12 lines

---

---
 mm/vmalloc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/vmalloc.c +0 -0
index a53c7462671bdd896f95712af71398ffbe22fb80..ff1876588b94ec69168324e93399dbd117a6959a 100644

mm/vmalloc.c static int vmap_range_noflush(unsigned long addr, unsigned long end, · line 308

Quoted 5 lines
mm/vmalloc.c +1 -1
@@ -305,7 +305,7 @@ static int vmap_range_noflush(unsigned long addr, unsigned long end,
 	int err;
 	pgtbl_mod_mask mask = 0;
 
-	might_sleep();
+	might_alloc(GFP_PGTABLE_KERNEL);

This will invariably add a might_sleep() and hence preserves the existing behaviour besides adding fs_reclaim_acquire() which is an improvement.

Quoted 9 lines
mm/vmalloc.c +0 -1
 	BUG_ON(addr >= end);
 
 	start = addr;
 
---

base-commit: ecc46e02e0abe025a6e840cba2d647f23fd1d721
change-id: 20251208-b4-vmalloc-might_alloc-754a791e4e10

Best regards,

Please add <asm-generic/pgalloc.h> in mm/vmalloc.c - otherwise it does not get built as GFP_PGTABLE_KERNEL is not available. But otherwise LGTM.

After fixing the build.

#2 lore raw

Hi Anshuman, thanks a lot for taking a look.

Quoted 2 lines · Anshuman Khandual

Please add <asm-generic/pgalloc.h> in mm/vmalloc.c - otherwise it does not get built as GFP_PGTABLE_KERNEL is not available. But otherwise LGTM.

Oh, but that's not correct, IIUC we shouldn't directly be including asm-generic headers from here.

So while in principle GFP_PGTABLE_KERNEL is a sensible common demoninator here, it doesn't actually exist at all everywhere, e.g. it doesn't look like m68k defines it for Motorola.

So maybe the best way here is a really vague:

/*
* Different archs allocate pagetables in different ways, assume
* GFP_KERNEL as a common denominator.
*/
might_alloc(GFP_KERNEL)

... a bit yucky but I think still useful.

Any thoughts?

maintainer
#3 lore raw
Quoted 21 lines · Brendan Jackman

Hi Anshuman, thanks a lot for taking a look.

Quoted 2 lines · Anshuman Khandual

Please add <asm-generic/pgalloc.h> in mm/vmalloc.c - otherwise it does not get built as GFP_PGTABLE_KERNEL is not available. But otherwise LGTM.

Oh, but that's not correct, IIUC we shouldn't directly be including asm-generic headers from here.

So while in principle GFP_PGTABLE_KERNEL is a sensible common demoninator here, it doesn't actually exist at all everywhere, e.g. it doesn't look like m68k defines it for Motorola.

So maybe the best way here is a really vague:

/*
* Different archs allocate pagetables in different ways, assume
* GFP_KERNEL as a common denominator.
*/
might_alloc(GFP_KERNEL)

Quoted 4 lines

... a bit yucky but I think still useful.

Any thoughts?

Maybe just add a comment why that might_sleep() is there?

We can simply state that page table allocation path is hard-coded internally to use GFP_KERNEL flag.

Thanks!

--
Uladzislau Rezki

syzbot Tested-by [email protected] on v1 1/1
#4 lore raw
In reply to Brendan Jackman
syzbot ci has tested the following series

syzbot ci has tested the following series

[v1] mm/vmalloc: clarify why vmap_range_noflush() might sleep
https://lore.kernel.org/all/[email protected]
* [PATCH] mm/vmalloc: clarify why vmap_range_noflush() might sleep

and found the following issue:
kernel build error

Full report is available here:
https://ci.syzbot.org/series/ca5fb0b1-49a4-4066-b78d-d144e56e0d2b

***

kernel build error

tree:      mm-new
URL:       https://kernel.googlesource.com/pub/scm/linux/kernel/git/akpm/mm.git
base:      3f43be96f919cc611dcb2a4e38dd464831f4513e
arch:      amd64
compiler:  Debian clang version 20.1.8 (++20250708063551+0c9f909b7976-1~exp1~20250708183702.136), Debian LLD 20.1.8
config:    https://ci.syzbot.org/builds/1d630f58-694e-4568-8ac0-62e02ea8a59d/config

mm/vmalloc.c:308:14: error: use of undeclared identifier 'GFP_PGTABLE_KERNEL'

***

If these findings have caused you to resend the series or submit a separate fix, please add the following tag to your commit message:

---
This report is generated by a bot. It may contain errors. syzbot ci engineers can be reached at [email protected].

#6 lore raw
In reply to Brendan Jackman
kernel test robot noticed the following build errors:

Hi Brendan,

kernel test robot noticed the following build errors:

[auto build test ERROR on ecc46e02e0abe025a6e840cba2d647f23fd1d721]

url:    https://github.com/intel-lab-lkp/linux/commits/Brendan-Jackman/mm-vmalloc-clarify-why-vmap_range_noflush-might-sleep/20251208-132202
base:   ecc46e02e0abe025a6e840cba2d647f23fd1d721
patch link:    https://lore.kernel.org/r/20251208-b4-vmalloc-might_alloc-v1-1-94a9bb8ecb08%40google.com
patch subject: [PATCH] mm/vmalloc: clarify why vmap_range_noflush() might sleep
config: nios2-allnoconfig (https://download.01.org/0day-ci/archive/20251209/[email protected]/config)
compiler: nios2-linux-gcc (GCC) 11.5.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251209/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

mm/vmalloc.c: In function 'vmap_range_noflush':

>> mm/vmalloc.c:308:21: error: 'GFP_PGTABLE_KERNEL' undeclared (first use in this function)
     308 |         might_alloc(GFP_PGTABLE_KERNEL);
         |                     ^~~~~~~~~~~~~~~~~~

mm/vmalloc.c:308:21: note: each undeclared identifier is reported only once for each function it appears in

vim +/GFP_PGTABLE_KERNEL +308 mm/vmalloc.c

297
298 static int vmap_range_noflush(unsigned long addr, unsigned long end,
299 phys_addr_t phys_addr, pgprot_t prot,
300 unsigned int max_page_shift)
301 {
302 pgd_t *pgd;
303 unsigned long start;
304 unsigned long next;
305 int err;
306 pgtbl_mod_mask mask = 0;
307

 > 308		might_alloc(GFP_PGTABLE_KERNEL);

309 BUG_ON(addr >= end);
310
311 start = addr;
312 pgd = pgd_offset_k(addr);
313 do {
314 next = pgd_addr_end(addr, end);
315 err = vmap_p4d_range(pgd, addr, next, phys_addr, prot,
316 max_page_shift, &mask);
317 if (err)
318 break;
319 } while (pgd++, phys_addr += (next - addr), addr = next, addr != end);
320
321 if (mask & ARCH_PAGE_TABLE_SYNC_MASK)
322 arch_sync_kernel_mappings(start, end);
323
324 return err;
325 }
326

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
#7 lore raw
In reply to Brendan Jackman
kernel test robot noticed the following build errors:

Hi Brendan,

kernel test robot noticed the following build errors:

[auto build test ERROR on ecc46e02e0abe025a6e840cba2d647f23fd1d721]

url:    https://github.com/intel-lab-lkp/linux/commits/Brendan-Jackman/mm-vmalloc-clarify-why-vmap_range_noflush-might-sleep/20251208-132202
base:   ecc46e02e0abe025a6e840cba2d647f23fd1d721
patch link:    https://lore.kernel.org/r/20251208-b4-vmalloc-might_alloc-v1-1-94a9bb8ecb08%40google.com
patch subject: [PATCH] mm/vmalloc: clarify why vmap_range_noflush() might sleep
config: x86_64-allnoconfig (https://download.01.org/0day-ci/archive/20251209/[email protected]/config)
compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251209/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

>> mm/vmalloc.c:308:14: error: use of undeclared identifier 'GFP_PGTABLE_KERNEL'
     308 |         might_alloc(GFP_PGTABLE_KERNEL);
         |                     ^

1 error generated.

vim +/GFP_PGTABLE_KERNEL +308 mm/vmalloc.c

297
298 static int vmap_range_noflush(unsigned long addr, unsigned long end,
299 phys_addr_t phys_addr, pgprot_t prot,
300 unsigned int max_page_shift)
301 {
302 pgd_t *pgd;
303 unsigned long start;
304 unsigned long next;
305 int err;
306 pgtbl_mod_mask mask = 0;
307

 > 308		might_alloc(GFP_PGTABLE_KERNEL);

309 BUG_ON(addr >= end);
310
311 start = addr;
312 pgd = pgd_offset_k(addr);
313 do {
314 next = pgd_addr_end(addr, end);
315 err = vmap_p4d_range(pgd, addr, next, phys_addr, prot,
316 max_page_shift, &mask);
317 if (err)
318 break;
319 } while (pgd++, phys_addr += (next - addr), addr = next, addr != end);
320
321 if (mask & ARCH_PAGE_TABLE_SYNC_MASK)
322 arch_sync_kernel_mappings(start, end);
323
324 return err;
325 }
326

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki