[Crash-utility] shared pending signal queue added to sig command

[Dave Anderson]

Olivier Daudel wrote:

> ----- Original Message -----
> From: Dave Anderson
> To: Discussion list for crash utility usage, maintenance and development
> Sent: Wednesday, September 13, 2006 10:57 PM
> Subject: Re: [Crash-utility] shared pending signal queue added to sig
> command
>
> Sorry -- I can't use it as us because it doesn't work with 2.4 kernels,
> or at least with RHEL3 kernels.  Here's a RHEL3 (2.4.21-37.ELsmp)
>
> I don't actualy have this Linux version, but in looking in the task.c code,
> i suppose
> OFFSET_OPTION(sigpending_head, sigpending_list)) should do the job ?
>
> Three other things:
>
> (1) Shouldn't your "SHARED_PENDING" line have a following
>     "yes" or "no" indication, as is the case with the
>     "SIGPENDING:" output?
>
> For me the test on TIF_SIGPENDING is where the decision is made.
>
> (2) Thanks for remembering to add "signal_struct_shared_pending"
>     do dump_offset_table() -- but also don't forget to add
>     the new "sigpending_signal" addition you made to the
>     size_table structure, which is also dumped inside the
>     dump_offset_table() function.
>
> OK.
>
> (3) The SIQUEUE output is now associated with either SIGPENDING
>     or SHARED_PENDING, so the two SIGQUEUE outputs should be
>     indented so it's more obvious.
>
> OK for that.
> Also, if we have both queues, i use PRIVATE_PENDING and SHARED_PENDING.
>
> SIGPENDING: no
> PRIVATE_PENDING
>        SIGNAL: 0000000200000800
>    BLOCKED: 8000000200000800
>   SIGQUEUE:  SIG  SIGINFO
>                           12  f4561a34
>                           34  f45619a0
> SHARED_PENDING
>        SIGNAL: 8000000000000800
>   SIGQUEUE:  SIG  SIGINFO
>                           12  f456190c
>                           64  f4561f68
>                           64  f4561ed4
>

The code seems to work on 2.4 OK now.

But can you update the sig "help" page to reflect the changes
you've made to the output format?  You can just send a patch
for help.c.

Thanks,
  Dave