Re: [PATCH] Limit the switch_to_thread() calls in startup_inferior()

[Simon Marchi <simark@simark.ca>]

On 2020-11-01 1:33 p.m., Samuel Thibault wrote:
> Hello,
>
> (Note that I have never really hacked very much into gdb's inferior
> management, so I don't know much how things are supposed to be working.)
>
> Simon Marchi, le sam. 24 oct. 2020 22:58:05 -0400, a ecrit:
>> I tried to wrap my head around this and understand the need for the
>> switch_to_thread, what could be the impacts of your change, and I just
>> can't give anything close to a definitive answer.
>>
> [...]
>> Samuel, as the last person who touched gnu-nat.c, do you have a bit of
>> time to test Kamil's patch on Hurd, see if it breaks starting inferiors?
>
> I am not really sure what I am supposed to test. I tried a program that
> starts a second thread which just pause()s. With Kamil's patch I can
> switch & backtrace between the two threads.

Just testing a "run" would have been sufficient, but you've tested even
more, which doesn't hurt.

>
> (gdb) r
> Starting program: /home/samy/test
> ^C[New Thread 13012.6]
> [New Thread 13012.7]
>
> Thread 5 received signal SIGINT, Interrupt.
> 0x0107951c in ?? () from /lib/i386-gnu/libc.so.0.3
> (gdb) info thread
>   Id   Target Id         Frame
> * 5    Thread 13012.5    0x0107951c in ?? () from /lib/i386-gnu/libc.so.0.3
>   6    Thread 13012.6    0x0109dc74 in _hurd_intr_rpc_mach_msg ()
>    from /lib/i386-gnu/libc.so.0.3
>   7    Thread 13012.7    0x0107951c in ?? () from /lib/i386-gnu/libc.so.0.3
> (gdb) bt
> #0  0x0107951c in ?? () from /lib/i386-gnu/libc.so.0.3
> #1  0x01079cc6 in mach_msg () from /lib/i386-gnu/libc.so.0.3
> [...]
> #5  0x0104cdca in pthread_join () from /lib/i386-gnu/libpthread.so.0.3
> #6  0x0803254b in main () at test.c:13
> (gdb) thread 7
> [Switching to thread 7 (Thread 13012.7)]
> #0  0x0107951c in ?? () from /lib/i386-gnu/libc.so.0.3
> (gdb) bt
> #0  0x0107951c in ?? () from /lib/i386-gnu/libc.so.0.3
> [...]
> #3  0x0114e4ce in pause () from /lib/i386-gnu/libc.so.0.3
> #4  0x080324fa in f (foo=0x0) at test.c:6
> #5  0x0104c3ff in entry_point () from /lib/i386-gnu/libpthread.so.0.3
> #6  0x00000000 in ?? ()

Ok, so it all seems good.

>> And what if you remove all the switch_to_thread from that function, does
>> it still work?
>
> If I remove all calls from that fonction I am getting
>
> (gdb) r
> Starting program: /home/samy/test
> Can't fetch registers from thread bogus thread id 1: No such thread
> (gdb) info thread
>   Id   Target Id         Frame
>   5    Thread 18821.5    0x000021e0 in ?? ()

> The current thread <Thread ID 1> has terminated.  See `help thread'.

Ok, so if we remove these switch_to_thread calls, we get back to the
initial problem reported in

    https://sourceware.org/legacy-ml/gdb/2008-10/msg00034.html

So we know they are still necessary, thanks for your help.

> Since I now have a built tree I can easily experiment more.

I think we have all we need for now, thanks!

Simon