Re: [PATCH 3/3] gdbserver/Windows: crash during connection establishment phase

[Pedro Alves <palves@redhat.com>]

On 05/04/2018 07:58 PM, Joel Brobecker wrote:
>>> +#ifndef IN_PROCESS_AGENT
>>> +#ifdef __x86_64__
>>> +  static const char *expedite_regs_amd64[] = { "rbp", "rsp", "rip", NULL };
>>> +  tdesc->expedite_regs = expedite_regs_amd64;
>>> +#else /* __x86_64__ */
>>> +  static const char *expedite_regs_i386[] = { "ebp", "esp", "eip", NULL };
>>> +  tdesc->expedite_regs = expedite_regs_i386;
>>> +#endif /* __x86_64__ */
>>> +#endif
>>
>> Won't all x86 ports have the same problem?  I.e., don't
>> nto-x86-low.c:nto_x86_arch_setup and
>> lynx-i386-low.c:lynx_i386_arch_setup need the same treatment?
>> Should we put those arrays in some shared i386 file?
> 
> They probably would, but these are platforms I cannot test at
> the moment, so I didn't want to take a risk and create a situation
> for them. I'm quite happy to adjust the way you suggest.

I think I'd take the risk anyway, even without testing.  (Just like
the original patch that broke them didn't get testing on !Linux
ports.)  They're already most probably broken, so we can't be
making it worse, IMO.

Thanks,
Pedro Alves