Re: Linux 5.19-rc7

[Guenter Roeck]

On Sun, Jul 17, 2022 at 02:02:37PM -0700, Linus Torvalds wrote:
> It's a Sunday afternoon, I wonder what that might mean..
> 
> Another week, another rc. We obviously had that whole "Retbleed"
> thing, and it does show up in both the diffstat and the shortlog, and
> rc7 is definitely bigger than usual.
> 
> And also as usual, when we've had one of those embargoed hw issues
> pending, the patches didn't get the open development, and then as a
> result missed all the usual sanity checking by all the automation
> build and test infrastructure we have. So no surprise - there's been
> various small fixup patches afterwards too for some corner cases.
> 
> That said, last week there were two other development trees that
> independently also asked for an extension, so 5.19 will be one of
> those releases that have an additional rc8 next weekend before the
> final release. We had some last-minute btrfs reverts, and there's also
> a pending issue with a intel GPU firmware.
> 
> When it rains it pours.
> 
> Not that things really look all that bad. I think we've got the
> retbleed fallout all handled (knock wood), and the btrfs reverts are
> in place. And the Intel GPU firmware issue seems to have a patch
> pending too (or we'll just revert). So it's not like we have any huge
> issues, but an extra week is most definitely called for.
> 

Build results look good.

Build results:
	total: 150 pass: 150 fail: 0
Qemu test results:
	total: 489 pass: 489 fail: 0

As for warnings, I see

[   19.798382] test_bitmap: parselist: 14: input is '0-2047:128/256' OK, Time: 12584
[   19.803177] ==================================================================
[   19.803628] BUG: KFENCE: out-of-bounds read in _find_next_bit_le+0x10/0x48
[   19.803628]
[   19.804846] Out-of-bounds read at 0xef5b2000 (4096B right of kfence-#103):
[   19.805333]  _find_next_bit_le+0x10/0x48
[   19.805561]
[   19.805782] kfence-#103: 0xef5b1000-0xef5b1fff, size=4096, cache=kmalloc-4k
[   19.805782]
[   19.806263] allocated by task 1 on cpu 1 at 19.800216s:
[   19.806797]  test_bitmap_printlist+0x2c/0x13c
[   19.807024]  test_bitmap_init+0x64/0x478
[   19.807197]  do_one_initcall+0x6c/0x3a4
[   19.807380]  kernel_init_freeable+0x1c4/0x230
[   19.807568]  kernel_init+0x18/0x130
[   19.807729]  ret_from_fork+0x14/0x2c
[   19.807885]  0x0
[   19.808520] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.19.0-rc7 #1
[   19.808924] Hardware name: Samsung Exynos (Flattened Device Tree)
[   19.809301] PC is at _find_next_bit_le+0x10/0x48
[   19.809516] LR is at bitmap_list_string.constprop.0+0xfc/0x144
[   19.809775] pc : [<c06ff0cc>]    lr : [<c0715870>]    psr: 20000113
[   19.810033] sp : f082dc70  ip : 00000001  fp : 00000001
[   19.810254] r10: 00000000  r9 : 0000002d  r8 : ef5b1000
[   19.810478] r7 : c0e55514  r6 : c2215000  r5 : 00008000  r4 : 00008000
[   19.810746] r3 : 3163db73  r2 : 00008001  r1 : 00008000  r0 : ef5b1000
[   19.811082] Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
[   19.811418] Control: 10c5387d  Table: 4000406a  DAC: 00000051
[   19.811791] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.19.0-rc7 #1
[   19.812049] Hardware name: Samsung Exynos (Flattened Device Tree)
[   19.812437]  unwind_backtrace from show_stack+0x10/0x14
[   19.812723]  show_stack from dump_stack_lvl+0x68/0x90
[   19.812953]  dump_stack_lvl from kfence_report_error+0x260/0x664
[   19.813231]  kfence_report_error from kfence_handle_page_fault+0x1ec/0x27c
[   19.813515]  kfence_handle_page_fault from __do_kernel_fault.part.0+0x3c/0x74
[   19.813799]  __do_kernel_fault.part.0 from do_page_fault+0x1d0/0x408
[   19.814070]  do_page_fault from do_DataAbort+0x3c/0xb4
[   19.814292]  do_DataAbort from __dabt_svc+0x50/0x80
[   19.814614] Exception stack(0xf082dc20 to 0xf082dc68)
[   19.815104] dc20: ef5b1000 00008000 00008001 3163db73 00008000 00008000 c2215000 c0e55514
[   19.815460] dc40: ef5b1000 0000002d 00000000 00000001 00000001 f082dc70 c0715870 c06ff0cc
[   19.815773] dc60: 20000113 ffffffff
[   19.815964]  __dabt_svc from _find_next_bit_le+0x10/0x48

which I don't recall seeing before. Digging into it, it may be spurious
(ie seen with various emulations and not with every boot), and it looks
like it is not not a new problem: I see the same or a similar report
in v5.18.y, but never in v5.15.y. I'll try to bisect.

Guenter