On 19.03.2020 16:30, Simon Marchi wrote:
> On 2020-03-19 8:55 a.m., Simon Marchi wrote:
>> On 2020-03-19 8:28 a.m., Kamil Rytarowski wrote:
>>> Unlike most other Operating Systems, NetBSD tracks both pid and lwp.
>>> The process id on NetBSD is stored always in the pid field of ptid.
>>>
>>> gdb/ChangeLog:
>>>
>>> 	* inf-ptrace.h: Disable get_ptrace_pid on NetBSD.
>>> 	* inf-ptrace.c: Likewise.
>>> 	* (gdb_ptrace): Add.
>>> 	* (inf_ptrace_target::resume): Update.
>>> 	* (inf_ptrace_target::xfer_partial): Likewise.
>>> 	* (inf_ptrace_peek_poke): Change argument `pid' to `ptid'.
>>> 	* (inf_ptrace_peek_poke): Update.
>>
>> Hmm, this breaks simple debugging on Linux:
>>
>> $ ./gdb --data-directory=data-directory a.out -ex start
>> Reading symbols from a.out...
>> Temporary breakpoint 1 at 0x4004da: file test.c, line 2.
>> Starting program: /home/smarchi/build/binutils-gdb/gdb/a.out
>>
>> Program received signal SIGILL, Illegal instruction.
>> 0x00007ffff7dda96d in dl_main (phdr=<optimized out>, phnum=<optimized out>, user_entry=<optimized out>, auxv=<optimized out>) at rtld.c:1517
>> 1517    rtld.c: No such file or directory.
>>
>>
>> I haven't figured out why by inspecting the code yet, I'll try to debug it later today.
>>
>> Simon
>>
> 
> Ah, it's because ptrace returns long and not int on GNU/Linux.  So when we want to read
> a 64-bits word from memory, it gets truncated.  gdb_ptrace should return PTRACE_TYPE_RET.
> 
> In fact, to be consistent, all these gdb_ptrace functions should be changed to return
> PTRACE_TYPE_RET (as a separate patch).
> 

Done in v5.

I will follow up with other gdb_ptrace() instances next.

> Simon
>