Skip to contentSkip to conversation

XArray: revert (unintentional?) behavior change

PATCH In review Tamir Duberstein 6 participants 7 messages Mar 22, 2025 – Mar 24, 2025

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 XArray: revert (unintentional?) behavior change

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

---
Before this change:

BUG at xa_alloc_index:57

  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.14.0-rc7-next-20250321-00001-gf45bb5d4b2e8 #854 NONE
  Hardware name: linux,dummy-virt (DT)
  Call trace:
   show_stack+0x18/0x24 (C)
   dump_stack_lvl+0x40/0x84
   dump_stack+0x18/0x24
   xa_alloc_index+0x118/0x15c
   check_xa_alloc_1+0x114/0x6e8
   check_xa_alloc+0x24/0x78
   xarray_checks+0x58/0xd4
   do_one_initcall+0x74/0x168
   do_initcall_level+0x8c/0xac
   do_initcalls+0x54/0x94
   do_basic_setup+0x18/0x24
   kernel_init_freeable+0xb8/0x120
   kernel_init+0x20/0x198
   ret_from_fork+0x10/0x20
  BUG at xa_alloc_index:57
  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.14.0-rc7-next-20250321-00001-gf45bb5d4b2e8 #854 NONE
  Hardware name: linux,dummy-virt (DT)
  Call trace:
   show_stack+0x18/0x24 (C)
   dump_stack_lvl+0x40/0x84
   dump_stack+0x18/0x24
   xa_alloc_index+0x118/0x15c
   check_xa_alloc_1+0x120/0x6e8
   check_xa_alloc+0x24/0x78
   xarray_checks+0x58/0xd4
   do_one_initcall+0x74/0x168
   do_initcall_level+0x8c/0xac
   do_initcalls+0x54/0x94
   do_basic_setup+0x18/0x24
   kernel_init_freeable+0xb8/0x120
   kernel_init+0x20/0x198
   ret_from_fork+0x10/0x20
  BUG at xa_erase_index:62
  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.14.0-rc7-next-20250321-00001-gf45bb5d4b2e8 #854 NONE
  Hardware name: linux,dummy-virt (DT)
  Call trace:
   show_stack+0x18/0x24 (C)
   dump_stack_lvl+0x40/0x84
   dump_stack+0x18/0x24
   xa_erase_index+0xb0/0xd8
   check_xa_alloc_1+0x12c/0x6e8
   check_xa_alloc+0x24/0x78
   xarray_checks+0x58/0xd4
   do_one_initcall+0x74/0x168
   do_initcall_level+0x8c/0xac
   do_initcalls+0x54/0x94
   do_basic_setup+0x18/0x24
   kernel_init_freeable+0xb8/0x120
   kernel_init+0x20/0x198
   ret_from_fork+0x10/0x20
  BUG at xa_erase_index:62
  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.14.0-rc7-next-20250321-00001-gf45bb5d4b2e8 #854 NONE
  Hardware name: linux,dummy-virt (DT)
  Call trace:
   show_stack+0x18/0x24 (C)
   dump_stack_lvl+0x40/0x84
   dump_stack+0x18/0x24
   xa_erase_index+0xb0/0xd8
   check_xa_alloc_1+0x160/0x6e8
   check_xa_alloc+0x24/0x78
   xarray_checks+0x58/0xd4
   do_one_initcall+0x74/0x168
   do_initcall_level+0x8c/0xac
   do_initcalls+0x54/0x94
   do_basic_setup+0x18/0x24
   kernel_init_freeable+0xb8/0x120
   kernel_init+0x20/0x198
   ret_from_fork+0x10/0x20
  XArray: 6782365 of 6782369 tests passed
---
 lib/xarray.c | 1 +
 1 file changed, 1 insertion(+)
lib/xarray.c +1 -0
index 261814d170d8..638c4c90e5b2 100644
@@ -2380,6 +2380,7 @@ void xa_destroy(struct xarray *xa)
 	unsigned long flags;
 	void *entry;
 
+	xas.xa_node = NULL;
 	xas_lock_irqsave(&xas, flags);
 	entry = xa_head_locked(xa);
 	RCU_INIT_POINTER(xa->xa_head, NULL);

---
base-commit: 9388ec571cb1adba59d1cded2300eeb11827679c
change-id: 20250321-xarray-fix-destroy-274090a7a18b

Best regards,

Signature · 1 line
Tamir Duberstein <[email protected]>
maintainer
#1 lore raw
Quoted 4 lines · On Fri, 21 Mar 2025 22:17:08 -0400 Tamir Duberstein <[email protected]> wrote:

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

Thanks.

6684aba0780d appears to be only in linux-next. It has no Link: and my efforts to google its origin story failed. Help?

#2 lore raw

Hi Andrew,

Quoted 6 lines · On Fri, 21 Mar 2025 21:37:33 -0700 Andrew Morton <[email protected]>…
Quoted 3 lines · On Fri, 21 Mar 2025 22:17:08 -0400 Tamir Duberstein <[email protected]> wrote:

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Quoted 6 lines
Quoted 1 line
Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

Thanks.

6684aba0780d appears to be only in linux-next. It has no Link: and my efforts to google its origin story failed. Help?

Its been sitting in Matthew's xarray tree
(git://git.infradead.org/users/willy/xarray.git#main) since Sep 24 2024 and linux-next since then.

Signature · 2 lines
Cheers,
Stephen Rothwell
#3 lore raw
In reply to Andrew Morton

Le 22/03/2025 à 05:37, Andrew Morton a écrit :

Quoted 5 lines · On Fri, 21 Mar 2025 22:17:08 -0400 Tamir Duberstein <[email protected]> wrote:
Quoted 4 lines

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Quoted 7 lines
Quoted 1 line
Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

Thanks.

6684aba0780d appears to be only in linux-next. It has no Link: and my efforts to google its origin story failed. Help?

Apparently comes from git://git.infradead.org/users/willy/xarray.git

#4 lore raw
In reply to Andrew Morton
Quoted 5 lines · Andrew Morton
Quoted 3 lines · On Fri, 21 Mar 2025 22:17:08 -0400 Tamir Duberstein <[email protected]> wrote:

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Quoted 7 lines
Quoted 1 line
Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

Thanks.

6684aba0780d appears to be only in linux-next. It has no Link: and my efforts to google its origin story failed. Help?

It was added back in september to willy's tree due to an XFS data corruption bug. See https://lore.kernel.org/linux-mm/[email protected]/ and
https://git.infradead.org/?p=users/willy/xarray.git;a=commitdiff;h=6684aba0780da9f505c202f27e68ee6d18c0aa66.

Signature · 1 line
Pedro
maintainer
#5 lore raw
In reply to Tamir Duberstein
Quoted 4 lines · Tamir Duberstein

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

This doesn't fix anything. The first failure is:

#6 0x0000555555649979 in XAS_INVALID (xas=xas@entry=0x7ffff4a003a0)

    at ../shared/linux/../../../../include/linux/xarray.h:1434
#7  0x000055555564f545 in check_xas_retry (xa=xa@entry=0x55555591ba00 <array>)
--Type <RET> for more, q to quit, c to continue without paging--
    at ../../../lib/test_xarray.c:131

#8 0x0000555555663869 in xarray_checks () at ../../../lib/test_xarray.c:2221 #9 0x00005555556639ab in xarray_tests () at xarray.c:15

That has nothing to do with xa_destroy(). What on earth are you doing?

Anyway, I'm at LSFMM and it'a Saturday. I shan't be looking at this until the 27th. There's clearly no urgency since you're the first one to notice in six months.

#6 lore raw
Quoted 9 lines · Matthew Wilcox
Quoted 4 lines · Tamir Duberstein

Partially revert commit 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends"), fixing test failures in check_xa_alloc.

Fixes 6684aba0780d ("XArray: Add extra debugging check to xas_lock and friends")

This doesn't fix anything. The first failure is:

Quoted 7 lines

#6 0x0000555555649979 in XAS_INVALID (xas=xas@entry=0x7ffff4a003a0)

    at ../shared/linux/../../../../include/linux/xarray.h:1434
#7  0x000055555564f545 in check_xas_retry (xa=xa@entry=0x55555591ba00 <array>)
--Type <RET> for more, q to quit, c to continue without paging--
    at ../../../lib/test_xarray.c:131

#8 0x0000555555663869 in xarray_checks () at ../../../lib/test_xarray.c:2221 #9 0x00005555556639ab in xarray_tests () at xarray.c:15

That's not what I see when I boot a kernel with CONFIG_TEST_XARRAY=y.

Quoted 1 line

That has nothing to do with xa_destroy(). What on earth are you doing?

I'm running the kernel in a VM on arm64. What are you doing?

Quoted 3 lines

Anyway, I'm at LSFMM and it'a Saturday. I shan't be looking at this until the 27th. There's clearly no urgency since you're the first one to notice in six months.

Sure. I misunderstood the purpose of linux-next, thinking that if a commit is in there then it will soon head to mainline. I realize now this isn't the case.