On 7/31/24 3:24 AM, Naresh Kamboju wrote:
On Tue, 30 Jul 2024 at 21:21, Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
This is the start of the stable review cycle for the 6.1.103 release. There are 440 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.
Responses should be made by Thu, 01 Aug 2024 15:14:54 +0000. Anything received after that time might be too late.
The whole patch series can be found in one patch at: https://www.kernel.org/pub/linux/kernel/v6.x/stable-review/patch-6.1.103-rc1... or in the git tree and branch at: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git linux-6.1.y and the diffstat can be found below.
thanks,
greg k-h
As others reported the boot failures and bisection results. I started validating 6.1.103-rc2 results.
[only for the record] The stable-rc 6.1 boot failed on all devices due to the following kernel panic.
Reported-by: Linux Kernel Functional Testing lkft@linaro.org
Boot crash log:
<1>[ 6.427635] Unable to handle kernel NULL pointer dereference at virtual address 00000000000000d0 <1>[ 6.428491] Mem abort info: <1>[ 6.428852] ESR = 0x0000000096000004 <1>[ 6.429458] EC = 0x25: DABT (current EL), IL = 32 bits <1>[ 6.430263] SET = 0, FnV = 0 <1>[ 6.430677] EA = 0, S1PTW = 0 <1>[ 6.431170] FSC = 0x04: level 0 translation fault <1>[ 6.431711] Data abort info: <1>[ 6.432167] ISV = 0, ISS = 0x00000004 <1>[ 6.432667] CM = 0, WnR = 0 <1>[ 6.433136] user pgtable: 4k pages, 48-bit VAs, pgdp=0000000100005000 <1>[ 6.433848] [00000000000000d0] pgd=0000000000000000, p4d=0000000000000000 <0>[ 6.435845] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP <4>[ 6.436795] Modules linked in: ip_tables x_tables <4>[ 6.437973] CPU: 1 PID: 1 Comm: systemd Tainted: G N 6.1.103-rc1 #1 <4>[ 6.438884] Hardware name: linux,dummy-virt (DT) <4>[ 6.439800] pstate: 02400009 (nzcv daif +PAN -UAO +TCO -DIT -SSBS BTYPE=--) <4>[ 6.440501] pc : dd_limit_depth (block/mq-deadline.c:609 block/mq-deadline.c:630) <4>[ 6.442475] lr : __blk_mq_alloc_requests (block/blk-mq.c:474) <4>[ 6.443041] sp : ffff80000802b7a0 <4>[ 6.443397] x29: ffff80000802b7b0 x28: 7fffffffffffffff x27: 0000000000000000 <4>[ 6.444406] x26: ffffa2ac0d3e5328 x25: ffff0000c0ad9cc0 x24: ffff0000c15986a0 <4>[ 6.445160] x23: 0000000000000001 x22: ffff0000c0ad9cc0 x21: ffff80000802ba38 <4>[ 6.445905] x20: ffff0000c0ad9cc0 x19: ffff80000802b820 x18: 0000000000000000 <4>[ 6.446654] x17: 0000000000000000 x16: 0000000000000000 x15: ffffffffffffffff <4>[ 6.447404] x14: ffffffffffffffff x13: ffffffffffffffff x12: ffffffffffffffff <4>[ 6.448159] x11: 0000000000000000 x10: 0000000000000000 x9 : ffffa2ac0c60feb0 <4>[ 6.448953] x8 : ffff80000802b668 x7 : 0000000000000000 x6 : 0000000000000001 <4>[ 6.449694] x5 : ffffa2ac0ea3f000 x4 : 0000000000000000 x3 : ffff0000c0f7bc00 <4>[ 6.450440] x2 : ffff0000c0ad9cc0 x1 : ffff80000802b820 x0 : 0000000000000000 <4>[ 6.451329] Call trace: <4>[ 6.451720] dd_limit_depth (block/mq-deadline.c:609 block/mq-deadline.c:630) <4>[ 6.452228] blk_mq_submit_bio (block/blk-mq.c:2911 block/blk-mq.c:3011) <4>[ 6.452712] __submit_bio (block/blk-core.c:595) <4>[ 6.453098] submit_bio_noacct_nocheck (include/linux/bio.h:614 block/blk-core.c:668 block/blk-core.c:696) <4>[ 6.453610] submit_bio_noacct (block/blk-core.c:799) <4>[ 6.454026] submit_bio (block/blk-core.c:828) <4>[ 6.454440] ext4_io_submit (fs/ext4/page-io.c:380) <4>[ 6.454884] ext4_writepages (fs/ext4/inode.c:2876) <4>[ 6.455328] do_writepages (mm/page-writeback.c:2491) <4>[ 6.455788] filemap_fdatawrite_wbc (mm/filemap.c:388 mm/filemap.c:378) <4>[ 6.456243] __filemap_fdatawrite_range (mm/filemap.c:422) <4>[ 6.456750] file_write_and_wait_range (mm/filemap.c:774) <4>[ 6.457249] ext4_sync_file (fs/ext4/fsync.c:151) <4>[ 6.457644] vfs_fsync_range (fs/sync.c:189) <4>[ 6.458095] do_fsync (fs/sync.c:202 fs/sync.c:212) <4>[ 6.458491] __arm64_sys_fsync (fs/sync.c:218) <4>[ 6.458899] invoke_syscall (arch/arm64/include/asm/current.h:19 arch/arm64/kernel/syscall.c:57) <4>[ 6.459349] el0_svc_common.constprop.0 (arch/arm64/include/asm/daifflags.h:28 arch/arm64/kernel/syscall.c:148) <4>[ 6.459864] do_el0_svc (arch/arm64/kernel/syscall.c:205) <4>[ 6.460272] el0_svc (arch/arm64/include/asm/daifflags.h:28 arch/arm64/kernel/entry-common.c:133 arch/arm64/kernel/entry-common.c:142 arch/arm64/kernel/entry-common.c:638) <4>[ 6.460644] el0t_64_sync_handler (arch/arm64/kernel/entry-common.c:656) <4>[ 6.461099] el0t_64_sync (arch/arm64/kernel/entry.S:585) <0>[ 6.461765] Code: f9400022 d50323bf f9401820 f9400443 (f9406802) All code ======== 0: f9400022 ldr x2, [x1] 4: d50323bf autiasp 8: f9401820 ldr x0, [x1, #48] c: f9400443 ldr x3, [x2, #8] 10:* f9406802 ldr x2, [x0, #208] <-- trapping instruction
Code starting with the faulting instruction
0: f9406802 ldr x2, [x0, #208] <4>[ 6.463039] ---[ end trace 0000000000000000 ]--- <4>[ 6.464924] ------------[ cut here ]------------ <4>[ 6.465459] WARNING: CPU: 1 PID: 1 at kernel/exit.c:816 do_exit (kernel/exit.c:816 (discriminator 1)) <4>[ 6.466537] Modules linked in: ip_tables x_tables <4>[ 6.467108] CPU: 1 PID: 1 Comm: systemd Tainted: G D N 6.1.103-rc1 #1 <4>[ 6.467792] Hardware name: linux,dummy-virt (DT) <4>[ 6.468226] pstate: 42400009 (nZcv daif +PAN -UAO +TCO -DIT -SSBS BTYPE=--) <4>[ 6.468843] pc : do_exit (kernel/exit.c:816 (discriminator 1)) <4>[ 6.469218] lr : do_exit (kernel/exit.c:816 (discriminator 1)) <4>[ 6.469612] sp : ffff80000802b370 <4>[ 6.469941] x29: ffff80000802b3a0 x28: ffffa2ac0dbe3f48 x27: ffffa2ac0dbe3f40 <4>[ 6.470683] x26: 0000000000000000 x25: 0000000000000001 x24: ffff80000802b472 <4>[ 6.471437] x23: 0000000000000001 x22: 000000000000000b x21: ffff0000c02c0000 <4>[ 6.472185] x20: ffff0000c02dc000 x19: ffff0000c02b8000 x18: 0000000000000006 <4>[ 6.472916] x17: 3030303030303030 x16: 3030303030303020 x15: ffff80000802af00 <4>[ 6.473652] x14: 0000000000000000 x13: 205d393330333634 x12: 2e36202020205b3e <4>[ 6.474386] x11: 3030303030206563 x10: ffffa2ac0eabb978 x9 : ffffa2ac0d379d20 <4>[ 6.475118] x8 : ffff80000802b278 x7 : 0000000000000000 x6 : 0000000000000001 <4>[ 6.475845] x5 : ffffa2ac0ea3f000 x4 : ffffa2ac0ea3f2e8 x3 : 0000000000000000 <4>[ 6.476576] x2 : ffff0000c02b8000 x1 : ffff0000c02b8000 x0 : ffff80000802ba38 <4>[ 6.477310] Call trace: <4>[ 6.477595] do_exit (kernel/exit.c:816 (discriminator 1)) <4>[ 6.477948] make_task_dead (arch/arm64/include/asm/atomic_ll_sc.h:95 (discriminator 2) arch/arm64/include/asm/atomic.h:52 (discriminator 2) include/linux/atomic/atomic-arch-fallback.h:440 (discriminator 2) include/linux/atomic/atomic-instrumented.h:199 (discriminator 2) kernel/exit.c:968 (discriminator 2)) <4>[ 6.478357] die (arch/arm64/kernel/traps.c:239) <4>[ 6.478697] die_kernel_fault (arch/arm64/mm/fault.c:307) <4>[ 6.479134] __do_kernel_fault (arch/arm64/mm/fault.c:404) <4>[ 6.479574] do_page_fault (arch/arm64/mm/fault.c:658) <4>[ 6.479962] do_translation_fault (arch/arm64/mm/fault.c:671) <4>[ 6.480411] do_mem_abort (arch/arm64/mm/fault.c:803 (discriminator 1)) <4>[ 6.480781] el1_abort (arch/arm64/include/asm/daifflags.h:28 arch/arm64/kernel/entry-common.c:368) <4>[ 6.481142] el1h_64_sync_handler (arch/arm64/kernel/entry-common.c:455) <4>[ 6.481588] el1h_64_sync (arch/arm64/kernel/entry.S:580) <4>[ 6.481958] dd_limit_depth (block/mq-deadline.c:609 block/mq-deadline.c:630) <4>[ 6.482355] blk_mq_submit_bio (block/blk-mq.c:2911 block/blk-mq.c:3011) <4>[ 6.482778] __submit_bio (block/blk-core.c:595) <4>[ 6.483184] submit_bio_noacct_nocheck (include/linux/bio.h:614 block/blk-core.c:668 block/blk-core.c:696) <4>[ 6.483650] submit_bio_noacct (block/blk-core.c:799) <4>[ 6.484098] submit_bio (block/blk-core.c:828) <4>[ 6.484454] ext4_io_submit (fs/ext4/page-io.c:380) <4>[ 6.484858] ext4_writepages (fs/ext4/inode.c:2876) <4>[ 6.485285] do_writepages (mm/page-writeback.c:2491) <4>[ 6.485670] filemap_fdatawrite_wbc (mm/filemap.c:388 mm/filemap.c:378) <4>[ 6.486122] __filemap_fdatawrite_range (mm/filemap.c:422) <4>[ 6.486587] file_write_and_wait_range (mm/filemap.c:774) <4>[ 6.487048] ext4_sync_file (fs/ext4/fsync.c:151) <4>[ 6.487463] vfs_fsync_range (fs/sync.c:189) <4>[ 6.487848] do_fsync (fs/sync.c:202 fs/sync.c:212) <4>[ 6.488223] __arm64_sys_fsync (fs/sync.c:218) <4>[ 6.488649] invoke_syscall (arch/arm64/include/asm/current.h:19 arch/arm64/kernel/syscall.c:57) <4>[ 6.489039] el0_svc_common.constprop.0 (arch/arm64/include/asm/daifflags.h:28 arch/arm64/kernel/syscall.c:148) <4>[ 6.489522] do_el0_svc (arch/arm64/kernel/syscall.c:205) <4>[ 6.489876] el0_svc (arch/arm64/include/asm/daifflags.h:28 arch/arm64/kernel/entry-common.c:133 arch/arm64/kernel/entry-common.c:142 arch/arm64/kernel/entry-common.c:638) <4>[ 6.490212] el0t_64_sync_handler (arch/arm64/kernel/entry-common.c:656) <4>[ 6.490682] el0t_64_sync (arch/arm64/kernel/entry.S:585) <4>[ 6.491191] ---[ end trace 0000000000000000 ]--- <0>[ 6.492871] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b <2>[ 6.493803] SMP: stopping secondary CPUs <0>[ 6.494951] Kernel Offset: 0x22ac04000000 from 0xffff800008000000 <0>[ 6.495536] PHYS_OFFSET: 0x40000000 <0>[ 6.495944] CPU features: 0x00000,000d3cbf,e69a773f <0>[ 6.496943] Memory Limit: none <0>[ 6.497664] ---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b ]---
Links to boot log:
- https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-6.1.y/build/v6.1.10...
- https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-6.1.y/build/v6.1.10...
- Config: https://storage.tuxsuite.com/public/linaro/lkft/builds/2jyP6yz35Go9caJsvrdNO...
- Build: https://storage.tuxsuite.com/public/linaro/lkft/builds/2jyP6yz35Go9caJsvrdNO...
metadata:
## Build
- kernel: 6.1.103-rc1
- git: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git
- git commit: a90fe3a941868870c281a880358b14d42f530b07
- git describe: v6.1.102-441-ga90fe3a94186
- test details:
https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-6.1.y/build/v6.1.10...
I built and booted 6.1.103-rc3 and didn't hit anything. Does it still trigger with that one? If yes, how do I reproduce this?
There are no deadline changes since 6.1.102, and the block side is just some integrity bits, which don't look suspicious. The other part this could potentially be is the sbitmap changes, but...