GDB 10.1 breaks Linux lx-symbols

GDB 10.1 breaks Linux lx-symbols

[Phillip Raffeck]

Hi,

the value subscript bug present in 10.1 and fixed in commit
5b56203a7cadd545b33713e98e274e582242e090 breaks, among others, the
lx-symbols command from the Linux kernels gdb scripts, as these rely on
iterating the sections of the loaded modules, which in turn uses array
subscripting.

As the bug fix has already made it into branch gdb-10-branch, are there
any plans to release a version 10.2 with bug fixes any time soon?

Note: I know, you're not responsible for the gdb scripts of the Linux
kernel, I just wanted give a real-world example of consequences stemming
from that bug.

Best regards
Phillip Raffeck

Re: GDB 10.1 breaks Linux lx-symbols

[Simon Marchi]

On 2021-01-27 7:20 a.m., Phillip Raffeck wrote:
> Hi,
> 
> the value subscript bug present in 10.1 and fixed in commit
> 5b56203a7cadd545b33713e98e274e582242e090 breaks, among others, the
> lx-symbols command from the Linux kernels gdb scripts, as these rely on
> iterating the sections of the loaded modules, which in turn uses array
> subscripting.
> 
> As the bug fix has already made it into branch gdb-10-branch, are there
> any plans to release a version 10.2 with bug fixes any time soon?
> 
> Note: I know, you're not responsible for the gdb scripts of the Linux
> kernel, I just wanted give a real-world example of consequences stemming
> from that bug.
> 
> Best regards
> Phillip Raffeck
> 

Hi!

It all depends on our release manager, Joel, but yes in theory GDB 10.2 is
coming eventually, I just can't tell when.

I have some more series I would like to propose for GDB 10, but they are
pending review.

[PATCH 00/13] DWARF 5 rnglists & loclists fixes (PR 26813)
https://sourceware.org/pipermail/gdb-patches/2021-January/175221.html

[PATCH 0/4] Fix CU expansion queue-related problems
https://sourceware.org/pipermail/gdb-patches/2020-November/173372.html

Simon

Re: GDB 10.1 breaks Linux lx-symbols

[Joel Brobecker]

Hello,

> > the value subscript bug present in 10.1 and fixed in commit
> > 5b56203a7cadd545b33713e98e274e582242e090 breaks, among others, the
> > lx-symbols command from the Linux kernels gdb scripts, as these rely on
> > iterating the sections of the loaded modules, which in turn uses array
> > subscripting.
> > 
> > As the bug fix has already made it into branch gdb-10-branch, are there
> > any plans to release a version 10.2 with bug fixes any time soon?
> > 
> > Note: I know, you're not responsible for the gdb scripts of the Linux
> > kernel, I just wanted give a real-world example of consequences stemming
> > from that bug.
> > 
> > Best regards
> > Phillip Raffeck
> > 
> 
> Hi!
> 
> It all depends on our release manager, Joel, but yes in theory GDB 10.2 is
> coming eventually, I just can't tell when.

This is timely. I was thinking of starting the discussions about
the 10.2 release early February. From there, I'm hoping we'll be able
to release before the end of February, but that depends on the number
of pending issues that we've identified and the speed at which
we resolve them.

> I have some more series I would like to propose for GDB 10, but they are
> pending review.
> 
> [PATCH 00/13] DWARF 5 rnglists & loclists fixes (PR 26813)
> https://sourceware.org/pipermail/gdb-patches/2021-January/175221.html
> 
> [PATCH 0/4] Fix CU expansion queue-related problems
> https://sourceware.org/pipermail/gdb-patches/2020-November/173372.html
> 
> Simon

-- 
Joel