[Crash-utility] gcore: Segmentation fault due to renaming of old_rsp symbol in kernel

[HATAYAMA Daisuke]

Eric,

Sorry for very delayed response.

I reviewed the patch in detail and I found that rsp_scratch is used
very temporarily at the entry point of system call and purpose of the
commit 263042e4630a85e856b4a8cd72f28dab33ef4741, triggering the
renaming of old_rsp to rsp_scratch, is to move the location of user
stack pointer from thread_struct::usersp to pt_regs saved at the
bottom of kernel stack.

So I fixed the issue so that gcore always refers to pt_regs for user
stack pointer if there's no usersp as a member of thread_struct.

I'll release a new tarball with other few fixes tomorrow.
Please wait for a while.

From: HATAYAMA Daisuke <d.hatayama at jp.fujitsu.com>
Subject: Re: [Crash-utility] gcore: Segmentation fault due to renaming of old_rsp symbol in kernel
Date: Tue, 8 Nov 2016 13:53:47 +0900

> Dave, Eric
> 
> From: Dave Anderson <anderson at redhat.com>
> Subject: Re: [Crash-utility] gcore: Segmentation fault due to renaming of old_rsp symbol in kernel
> Date: Tue, 1 Nov 2016 10:29:28 -0400
> 
>> 
>> 
>> ----- Original Message -----
>>> > From: crash-utility-bounces at redhat.com [crash-utility-bounces at redhat.com]
>>> > on behalf of Dave Anderson [anderson at redhat.com]
>>> > Sent: Monday, October 31, 2016 4:47 PM
>>> 
>>> > It's always appreciated when bug reports come with proposed fixes, and
>>> > and your patch certainly looks reasonable to me. But the gcore extension
>>> > module is maintained by Daisuke Hatayama, and any changes will require
>>> > his ACK and a subsequent package update. Daisuke is a member of this
>>> > mailing list, but just to make sure he sees this, I've cc'd him directly
>>> > as well.
>>> 
>>> Oh, absolutely -- I expected that.  The message started as a plea for help
>>> which I didn't want to bother him directly with (but I knew he monitored
>>> this list).  Then as I did my due diligence, it morphed into a solution.  I
>>> should have switched gears and sent it directly to him at that point (and
>>> thank you for doing so) but I guess I was already on track to post it to the
>>> list.  Thanks for the feedback.
>> 
>> Just to clarify -- I didn't mean to imply that extension module bug reports and
>> patches shouldn't be posted to this list.  By all means they should be.
>> 
> 
> Sorry for delayed response.
> I'll review the patch this week.
> Please wait for a while.
> 
> --
> Thanks.
> HATAYAMA, Daisuke
> 
> --
> Crash-utility mailing list
> Crash-utility at redhat.com
> https://www.redhat.com/mailman/listinfo/crash-utility
--
Thanks.
HATAYAMA, Daisuke