Re: [PATCH v19 00/14] arm64: entry: Convert to Generic Entry

[Jinjie Ruan]

在 2026/9/25 8:05, Kees Cook 写道:
> On Thu, Sep 24, 2026 at 05:32:14PM +0800, Jinjie Ruan wrote:
>> 在 2026/9/23 2:29, Kees Cook 写道:
>> I tested the rr test cases on a Kunpeng HIP09 (with minor adaptations,
>> since Kunpeng CPUs aren't supported by default), and the results showed
>> no regressions.
> 
> Great! Thank you for getting that set up.
> 
>>
>> |               |    7.3-rc4 baseline     |   Patched                 |
>>
>> |  ------------ |------------------------ | ------------------------- |
>> |   result      |  87% passed, 190 failed |   88% passed, 186 failed  |
>>
>>
>> | Failure Mode   | 7.3-rc4 baseline |   Patched  |
>> | ------------   | ---------------- | ---------- |
>> | TICK_MISMATCH  |         76       |   74       |
>> | TIMEOUT        |         68       |   65       |
>> | GDB_SCRIPT     |         34       |   37       |
>> | PERF_EVENT_OPEN|         3        |   2        |
>> |   OVERSHOOT    |         1        |   0        |
>> |    OTHER       |         8        |   8        |
>> | -------------- |  --------------- | ---------- |
>> |  Total         |      190         |   186      |
> 
> GDB_SCRIPT failures went up? Are the tests just flaky?

Yes, you are right!

Those three tests failed in the baseline too — they just surfaced as
tick mismatches then and as GDB script timeouts now, so the
bucket boundary shifts with the race. That's flakiness, not a regression.

|     changes         |               testcase              |
|  -----------------  |  ---------------------------------- |
| OVERSHOOT -> TICK   | clone_interruption                  |
| TICK -> GDB_SCRIPT  | checkpoint_prctl_name               |
| TICK -> GDB_SCRIPT  | checkpoint_prctl_name-no-syscallbuf |
| TICK -> GDB_SCRIPT  | clone_interruption-no-syscallbuf    |
| PERF_OPEN -> TIMEOUT| clone_newflags                      |
| PASS ->  TIMEOUT    | pid_ns_shutdown-no-syscallbuf       |
| TIMEOUT -> PASS     | fstatat-no-syscallbuf               |
| TIMEOUT -> PASS     | futex_exit_race                     |
| TIMEOUT -> PASS     | futex_exit_race-no-syscallbuf       |
| TIMEOUT -> PASS     | futex_exit_race_sigsegv             |
| TIMEOUT -> PASS     | futex_exit_race_sigsegv-no-syscallbuf|

> 
>>> - MTE: I didn't see MTE in the cover letter's test list, and it seems
>>>   relevant since patch 13 puts _TIF_MTE_ASYNC_FAULT into
>>>   ARCH_EXIT_TO_USER_MODE_WORK, so an async tag check fault lands
>>>   on the reworked exit path. The MTE selftests showed no differences,
>>>   but a third of them don't actually run for me under QEMU, so they'd
>>>   be worth running on your hardware. (Testing MTE under QEMU saw
>>
>> Turns out our arm64 machines don't support MTE; I was testing it on QEMU.
> 
> Ah, dang. I wonder if we can find anyone with MTE hardware to test this.
> Any volunteers? :)
> 
> -Kees
> 

-- 
Best regards,
Jinjie