[PATCH v3 31/41] powerpc/32: Dismantle EXC_XFER_STD/LITE/TEMPLATE

Sun Aug 1 02:40:59 AEST 2021

Finn Thain <fthain at linux-m68k.org> a écrit :

> Hi Christophe,

> We found that reducing memory with the kernel parameter 'mem=464M' would
> prevent the crash.
>
> Below are Stan's notes from the 'git bisect' run. They include an
> additional failure mode that might be of interest. We've also observed
> "Kernel attempted to write user page (c6207c) - exploit attempt?" and
> "kernel BUG at arch/powerpc/kernel/interrupt.c:49!".
>

Interesting.  Can you provide detailed log of thoose two BUGs ?

Thanks
Christophe

>
> On Thu, 29 Jul 2021, Stanley J. Johnson wrote:
>
>> Here are the git bisect results (I saved the kernels and dmesg outputs
>> in case they're needed later, though it's easy enough to re-create the
>> results).
>>
>> We'll have up to two boots (only one if the first boot fails) for each
>> kernel on WS-1 with the default memory setting of 512M.
>>
>> 1) 5.12.0-pmac-08447-g85f3f17b5db (good)
>> -> works: no dmesg errs, X works
>> -> works: no dmesg errs, X works
>>
>> 2) 5.12.0-pmac-12750-g23243c1ace9 (bad)
>> (Crash and burn. I was tempted to mark this one "skip", since the errors
>> are different, though we've seen them in earlier tests. So there could
>> be multiple regressions. But let's keep chasing this one. I can re-run
>> the bisect later and skip some failures if necessary.)
>> -> fails: "Unable to contact settings server" after wdm login; dmesg errors:
>> ...
>> [  127.169294] BUG: Unable to handle kernel instruction fetch (NULL  
>> pointer?)
>> [  127.169397] Faulting instruction address: 0x00000000
>> [  127.169635] Oops: Kernel access of bad area, sig: 11 [#2]
>> [  127.169661] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [  127.169696] Modules linked in:
>> [  127.169736] CPU: 0 PID: 1877 Comm: gdbus Tainted: G      D  
>> 5.12.0-pmac-12750-g23243c1ace9 #9
>> [  127.169776] NIP:  00000000 LR: 00000000 CTR: 00000000
>> ...
>>
>> 3) 5.12.0-rc7-pmac-02335-gaeacb52a8de (good)
>> -> works: no dmesg errs, X works
>> -> works: no dmesg errs, X works
>>
>> 4) 5.12.0-pmac-11585-g95275402f66 (bad)
>> -> fails: dmesg errs, wdm login fails with "Unable to contact settings
>> server..."
>> ...
>> [   58.290883] BUG: Unable to handle kernel instruction fetch
>> [   58.291009] Faulting instruction address: 0x7cc903a4
>> [   58.291240] Oops: Kernel access of bad area, sig: 11 [#1]
>> [   58.291268] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [   58.291305] Modules linked in:
>> [   58.291349] CPU: 0 PID: 1750 Comm: Xorg Not tainted  
>> 5.12.0-pmac-11585-g95275402f66 #11
>> [   58.291391] NIP:  7cc903a4 LR: 7cc903a6 CTR: c0127eb8
>> [   58.291417] REGS: e2da1e70 TRAP: 0400   Not tainted  
>> (5.12.0-pmac-11585-g95275402f66)
>> [   58.291450] MSR:  40001030 <ME,IR,DR>  CR: 28008228  XER: 20000000
>> ...
>>
>> 5) 5.12.0-pmac-11068-g9d31d233895 (good)
>> -> works: no dmesg errs, X works
>> -> works: no dmesg errs, X works
>>
>> 6) 5.12.0-rc3-pmac-00258-ga9d2f9bb225 (bad)
>> -> fails: wdm dies at login, screen hangs, dmesg errs that I haven't
>> seen before:
>> ...
>> [   59.988732] InputThread[1751]: bad frame in sys_rt_sigreturn:  
>> 51a79ab6 nip 001023c8 lr 001023c0
>> [   62.079043] InputThread[1760]: bad frame in sys_rt_sigreturn:  
>> a98be235 nip 001023c8 lr 001023c0
>> [   64.348063] InputThread[1776]: bad frame in sys_rt_sigreturn:  
>> a46b945e nip 001023c8 lr 001023c0
>> [   66.413980] InputThread[1785]: bad frame in sys_rt_sigreturn:  
>> a98be235 nip 001023c8 lr 001023c0
>> [   68.486768] InputThread[1794]: bad frame in sys_rt_sigreturn:  
>> 677e693b nip 001023c8 lr 001023c0
>>
>> 7) 5.12.0-rc3-pmac-00129-g036fc2cb1dc (bad)
>> -> fails: dmesg errs, wdm hangs (wdm did not crash, screen seems hung)
>> ...
>> [  101.136941] BUG: Unable to handle kernel instruction fetch
>> [  101.137059] Faulting instruction address: 0xfea31f74
>> [  101.137296] Oops: Kernel access of bad area, sig: 11 [#1]
>> [  101.137323] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [  101.137358] Modules linked in:
>> [  101.137401] CPU: 0 PID: 1882 Comm: xfce4-session Not tainted  
>> 5.12.0-rc3-pmac-00129-g036fc2cb1dc #14
>> [  101.137443] NIP:  fea31f74 LR: fea31f74 CTR: c00cfef4
>> [  101.137470] REGS: e2e21e70 TRAP: 0400   Not tainted  
>> (5.12.0-rc3-pmac-00129-g036fc2cb1dc)
>> [  101.137502] MSR:  40001030 <ME,IR,DR>  CR: 220084e8  XER: 00000000
>> ...
>>
>> 8) 5.12.0-rc3-pmac-00064-g719e7e212c7 (good)
>> -> works: no dmesg errs, X works
>> -> works: no dmesg errs, X works
>>
>> 9) 5.12.0-rc3-pmac-00096-ga2308836880 (bad)
>> -> fails: dmesg errs, wdm hangs while validating login
>> ...
>> [  104.559951] BUG: Unable to handle kernel instruction fetch
>> [  104.560086] Faulting instruction address: 0xfe988ba4
>> [  104.560307] Oops: Kernel access of bad area, sig: 11 [#1]
>> [  104.560337] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [  104.560371] Modules linked in:
>> [  104.560415] CPU: 0 PID: 1834 Comm: xfce4-session Not tainted  
>> 5.12.0-rc3-pmac-00096-ga2308836880 #16
>> [  104.560457] NIP:  fe988ba4 LR: fe988ba4 CTR: c00cfee8
>> [  104.560484] REGS: e2d9de70 TRAP: 0400   Not tainted  
>> (5.12.0-rc3-pmac-00096-ga2308836880)
>> [  104.560517] MSR:  40001030 <ME,IR,DR>  CR: 220484e8  XER: 00000000
>> ...
>> [  114.371853] BUG: Unable to handle kernel instruction fetch
>> [  114.371998] Faulting instruction address: 0x3e62a270
>> [  114.372270] Oops: Kernel access of bad area, sig: 11 [#2]
>> [  114.372298] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [  114.372334] Modules linked in:
>> [  114.372382] CPU: 0 PID: 1858 Comm: xfwm4 Tainted: G      D  
>> 5.12.0-rc3-pmac-00096-ga2308836880 #16
>> [  114.372425] NIP:  3e62a270 LR: 3e62a271 CTR: c065eebc
>> [  114.372452] REGS: e2dede70 TRAP: 0400   Tainted: G      D  
>> (5.12.0-rc3-pmac-00096-ga2308836880)
>> [  114.372486] MSR:  40001030 <ME,IR,DR>  CR: 22002284  XER: 00000000
>> ...
>>
>> 10) 5.12.0-rc3-pmac-00080-g7a7d744ffe8 (bad)
>> -> fails: dmesg errs, "Unable to contact settings server..."
>> ...
>> [  170.622408] BUG: Unable to handle kernel instruction fetch
>> [  170.622546] Faulting instruction address: 0xfe9255d4
>> [  170.622777] Oops: Kernel access of bad area, sig: 11 [#1]
>> [  170.622806] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [  170.622841] Modules linked in:
>> [  170.622885] CPU: 0 PID: 1863 Comm: xfce4-session Not tainted  
>> 5.12.0-rc3-pmac-00080-g7a7d744ffe8 #17
>> [  170.622927] NIP:  fe9255d4 LR: fe9255d4 CTR: c00cfee8
>> [  170.622954] REGS: e2da1e70 TRAP: 0400   Not tainted  
>> (5.12.0-rc3-pmac-00080-g7a7d744ffe8)
>> [  170.622987] MSR:  40001030 <ME,IR,DR>  CR: 220484e8  XER: 00000000
>> ...
>> [  172.611235] BUG: Unable to handle kernel instruction fetch (NULL  
>> pointer?)
>> [  172.611364] Faulting instruction address: 0x00000000
>> [  172.611662] Oops: Kernel access of bad area, sig: 11 [#2]
>> [  172.611690] BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 PowerMac
>> [  172.611725] Modules linked in:
>> [  172.611773] CPU: 0 PID: 1871 Comm: gmain Tainted: G      D  
>> 5.12.0-rc3-pmac-00080-g7a7d744ffe8 #17
>> [  172.611814] NIP:  00000000 LR: 00000000 CTR: 00000000
>> [  172.611840] REGS: e2db9e70 TRAP: 0400   Tainted: G      D  
>> (5.12.0-rc3-pmac-00080-g7a7d744ffe8)
>> [  172.611874] MSR:  40001030 <ME,IR,DR>  CR: 28008468  XER: 00000000
>> ...
>>
>> 11) 5.12.0-rc3-pmac-00072-ga2b3e09ae41 (bad)
>> -> fails: kernel panic, I'd have to capture the output via a serial console
>>
>> 12) 5.12.0-rc3-pmac-00068-gacc142b6230 (bad)
>> -> fails: kernel panic
>>
>> 13) 5.12.0-rc3-pmac-00066-g8f6ff5bd9b7 (good)
>> -> works: no dmesg errs, X works
>> -> works: no dmesg errs, X works
>>
>> 14) 5.12.0-rc3-pmac-00067-g4c0104a83fc (bad)
>> -> fails: kernel panic
>>
>> The git bisect thinks this is the bad commit, but I'm not sure it means
>> anything at all:
>>
>> -----
>>
>> 4c0104a83fc3990a76a01a2f4e504251fa9814c4 is the first bad commit