Re: server.el test failures (was: Re: bug#9800: Incomplete truncated file buffers from the /proc filesystem)

[Jim Porter <jporterbugs@gmail.com>]

On 2/23/2023 3:42 PM, Paul Eggert wrote:
> On 2/15/23 10:20, Paul Eggert wrote:
>> On 2023-02-14 23:06, Jim Porter wrote:
>>> Interesting. Running "./configure && make -j4 && make -j4 check" on 
>>> both Emacs master and the 29 branch shows these tests passing for me.
>>
>> I tried running it again on Ubuntu, several times, and this time 
>> server-tests.el succeeded. As you say, it may be a timing-related bug.
> 
> I tried running it on Fedora 37 x86-64 on a slow machine (AMD Phenom II 
> X4 910e) and reproduced the timing bug. The output of 'make check' 
> contained:
> 
>      GEN      lisp/server-tests.log
>    Running 7 tests (2023-02-23 09:45:51-0800, selector `(not (or (tag 
> :expensive-test) (tag :unstable) (tag :nativecomp)))')
>    make[3]: *** [Makefile:174: lisp/server-tests.log] Error 157
> 
> I guess "Error 157" means signal 29 (157 - 128), which is SIGIO on this 
> platform.
> 
> This was with Emacs master (commit 
> 6411a9af03a4eb1a82db47a9642b11ba7edaaaf0).

Since the last message I posted, I'm now also seeing this test fail, 
though I get a segfault instead. I bisected this to commit 
a555abc56d5270cebe94f904189526d7ac433a94 ("Fix order of faces in 
'face-list'").

I'm pretty surprised by this, since that patch is *very* simple, but I 
can reliably segfault with it, and never segfault without it. I'll keep 
digging to see what's going on here.