[v2] gdb/MicroBlaze: Add support for native linux gdb

  namaskaaram

Attached is the updated patch. All the changes suggested on PATCH v1
have been made and tested.

I did not remove the (sal.line != 0) check as it seemed trivial.
However, I can make that a separate patch if necessary.

Our sysadm In the process of setting up a git compatible email - the
process may take more time; until then appreciate your patience with
the attachments.

Important Note:

1) gdb appears to have an elaborate scheme to handle C++ exceptions
and "tunnel" them through exception unaware C code. This code is with
readline. Anytime
an error is discovered (even a syntax error) by a module,
gdb_exception_error() thrown is tunneled through readline using the
TRY_SJLJ/CATCH_SJLJ macros
implemented using setjmp/longjmp.

2) There is an issue that I came across while testing; native gdb goes
into an infinite loop when handling errors. I analyzed it and found it
to be an obvious glibc
sysdeps/unix/sysv/linux/microblaze/____longjmp_chk.S issue and
reported it to AMD. It is possible I missed out a glibc patch; At this
time - this is not a gdb issue.

3) Plan to follow through [2] and ensure a gcc/glibc defect report is
created if needed.

dhanyavaadaaha
gopi

On Tue, Jun 2, 2026 at 5:58 AM Michael Eager <eager@eagerm.com> wrote:
>
> Please post an updated patch for review.
>
> On 5/31/26 11:38 PM, Gopi Kumar Bulusu wrote:
> > On Fri, May 29, 2026 at 8:26 PM Tom Tromey <tom@tromey.com> wrote:
> >>
> >>>>>>> Gopi Kumar Bulusu <gopi@sankhya.com> writes:
> >>
> >>> Attached is a patch to add native linux support for gdb including
> >>> cache target support..
> >>
> >> Thanks for the patch.
> >>
> >> I read through it and have some nits.  There's nothing very serious
> >> except this:
> >>
> >>> The original sources for the patch come from Xilinx/AMD Yocto git
> >>> repository (2025.2)
> >>
> >> Who wrote them and are they covered by the copyright assignment?
> >>
> >> This is the most important consideration, the patch can't land without
> >> this being clear.
> >
> > This work is being performed under a contract with Xilinx/AMD as
> > confirmed in a different response to
> > Simon and therefore covered by the copyright assignment.
> >
> > I will make the changes as described below and retest.
> >
> > Approved for commit ?
> >
> > dhanyavaadaaha
> > gopi
> >
> >>
> >>>  From a3328b4cdfc58783b5f290f22255b36e26a6fb14 Mon Sep 17 00:00:00 2001
> >>> From: Gopi Kumar Bulusu <gopi@sankhya.com>
> >>> Date: Thu, 7 May 2026 17:47:40 +0530
> >>> Subject: [PATCH] gdb/MicroBlaze: Add support for native linux gdb
> >>
> >> It's best to just git send-email rather than using an attachment.
> >>
> >>>      bfd/ChangeLog:
> >>
> >>>      * elf32-microblaze.c (microblaze_elf_grok_prstatus): New function.
> >>>        (microblaze_elf_grok_psinfo): Likewise.
> >>
> >>>      gdb/ChangeLog:
> >>
> >>>      * Makefile.in (HFILES_NO_SRCDIR): Add microblaze-linux-tdep.h
> >>>        ALLDEPFILES: Add microblaze-linux-nat.c
> >>
> >> gdb doesn't use ChangeLogs any more and this text should all be removed.
> >>
> >>> +#define MICROBLAZE_TARGET_HAS_GETREGS 0
> >>> +#define MICROBLAZE_TARGET_HAS_SETREGS 0
> >>
> >> This seems weird:
> >>
> >>> +/* Wrapper function around ptrace.  */
> >>> +
> >>> +static void
> >>> +fetch_target_gp_regs (int tid, elf_gregset_t *gregs)
> >>> +{
> >>> +  elf_greg_t *gregp = *gregs;
> >>> +
> >>> +#if MICROBLAZE_TARGET_HAS_GETREGS
> >>> +  if (ptrace (PTRACE_GETREGS, tid, 0, (long) gregp) < 0)
> >>> +    {
> >>> +      perror_with_name (_("Couldn't get registers"));
> >>> +      return;
> >>> +    }
> >>> +#error "this configuration did not work during testing"
> >>
> >> Normally we don't put new dead code into gdb.
> >>
> >> I think this and the defines could just be replaced with a comment
> >> explaining that PTRACE_GETREGS doesn't work on this platform.
> >>
> >>> +++ b/gdb/microblaze-linux-tdep.h
> >>> @@ -0,0 +1,24 @@
> >>> +/* Target-dependent code for GNU/Linux on MicroBlaze.
> >>> +
> >>> +   Copyright (C) 2021-2026 Free Software Foundation, Inc.
> >>> +
> >>> +   This file is part of GDB.
> >>> +
> >>> +   This program is free software; you can redistribute it and/or modify
> >>> +   it under the terms of the GNU General Public License as published by
> >>> +   the Free Software Foundation; either version 3 of the License, or
> >>> +   (at your option) any later version.
> >>> +
> >>> +   This program is distributed in the hope that it will be useful,
> >>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> >>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> >>> +   GNU General Public License for more details.
> >>> +
> >>> +   You should have received a copy of the GNU General Public License
> >>> +   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
> >>> +#ifndef MICROBLAZE_LINUX_TDEP_H
> >>> +#define MICROBLAZE_LINUX_TDEP_H
> >>> + /* Target descriptions.  */
> >>> + extern struct target_desc *tdesc_microblaze_linux;
> >>
> >> Blank line between the #define and the comment.
> >>
> >> Also this define should have a slightly different name, see
> >> check-include-guards.py.
> >>
> >>>          non_stack_instruction_found = 0;
> >>> +          cache->register_offsets[rd] = -imm;
> >>>          continue;
> >>
> >> Indentation looks wrong, probably not using tabs.
> >>
> >>> @@ -399,8 +400,7 @@ microblaze_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR start_pc)
> >>>       {
> >>>         sal = find_sal_for_pc (func_start, 0);
> >>
> >>> -      if (sal.end < func_end
> >>> -       && start_pc <= sal.end)
> >>> +      if (sal.line != 0 && sal.end <= func_end  && start_pc <= sal.end)
> >>>        start_pc = sal.end;
> >>
> >> No objection from me but this seems somewhat unrelated.
> >>
> >>
> >>>     /* Call for side effects.  */
> >>> -  get_frame_func (next_frame);
> >>> +  cache->pc = get_frame_func (next_frame);
> >>
> >> Please remove that comment as it is now wrong.
> >>
> >>> +  if (regnum == MICROBLAZE_SP_REGNUM)
> >>> +        regnum = 1;
> >>
> >> Wrong indentation.
> >>
> >>> +  if (regnum == -1) {
> >>> +    int i;
> >>> +
> >>> +    for (i = 0; i < MICROBLAZE_REDR_REGNUM; i++) {
> >>> +      regcache->raw_supply (i, regs + i * MICROBLAZE_REGISTER_SIZE);
> >>> +    }
> >>
> >> Wrong brace placement in this hunk.
> >>
> >>> +  cb (".reg", tdep->sizeof_gregset, tdep->sizeof_gregset, tdep->gregset, NULL,
> >>> +      cb_data);
> >>
> >> Normally in new code we're using 'nullptr' now.  Best would be to go
> >> through the whole patch and check for this.  You don't have to fix
> >> pre-existing code that is using NULL though.
> >>
> >>>     gdbarch *gdbarch
> >>>       = gdbarch_alloc (&info, gdbarch_tdep_up (new microblaze_gdbarch_tdep));
> >>
> >>> +  microblaze_gdbarch_tdep *tdep
> >>> +    = gdbarch_tdep<microblaze_gdbarch_tdep> (gdbarch);
> >>> +
> >>> +  tdep->gregset = NULL;
> >>> +  tdep->sizeof_gregset = 0;
> >>> +  tdep->fpregset = NULL;
> >>> +  tdep->sizeof_fpregset = 0;
> >>
> >> Here it would be better to just add inline initializers to the new
> >> fields in microblaze_gdbarch_tdep.  Then they'll automatically be
> >> initialized by the 'new'.
> >>
> >> Tom
> >
>
  

Hi Gopi --

I (or more accurately, git) fixed several whitespace issues.

Since this is a revised patch, has it been retested?


On 6/1/26 10:05 PM, Gopi Kumar Bulusu wrote:
> namaskaaram
> 
> Attached is the updated patch. All the changes suggested on PATCH v1
> have been made and tested.
> 
> I did not remove the (sal.line != 0) check as it seemed trivial.
> However, I can make that a separate patch if necessary.
> 
> Our sysadm In the process of setting up a git compatible email - the
> process may take more time; until then appreciate your patience with
> the attachments.
> 
> Important Note:
> 
> 1) gdb appears to have an elaborate scheme to handle C++ exceptions
> and "tunnel" them through exception unaware C code. This code is with
> readline. Anytime
> an error is discovered (even a syntax error) by a module,
> gdb_exception_error() thrown is tunneled through readline using the
> TRY_SJLJ/CATCH_SJLJ macros
> implemented using setjmp/longjmp.
> 
> 2) There is an issue that I came across while testing; native gdb goes
> into an infinite loop when handling errors. I analyzed it and found it
> to be an obvious glibc
> sysdeps/unix/sysv/linux/microblaze/____longjmp_chk.S issue and
> reported it to AMD. It is possible I missed out a glibc patch; At this
> time - this is not a gdb issue.
> 
> 3) Plan to follow through [2] and ensure a gcc/glibc defect report is
> created if needed.
> 
> dhanyavaadaaha
> gopi
> 
> On Tue, Jun 2, 2026 at 5:58 AM Michael Eager <eager@eagerm.com> wrote:
>>
>> Please post an updated patch for review.
>>
>> On 5/31/26 11:38 PM, Gopi Kumar Bulusu wrote:
>>> On Fri, May 29, 2026 at 8:26 PM Tom Tromey <tom@tromey.com> wrote:
>>>>
>>>>>>>>> Gopi Kumar Bulusu <gopi@sankhya.com> writes:
>>>>
>>>>> Attached is a patch to add native linux support for gdb including
>>>>> cache target support..
>>>>
>>>> Thanks for the patch.
>>>>
>>>> I read through it and have some nits.  There's nothing very serious
>>>> except this:
>>>>
>>>>> The original sources for the patch come from Xilinx/AMD Yocto git
>>>>> repository (2025.2)
>>>>
>>>> Who wrote them and are they covered by the copyright assignment?
>>>>
>>>> This is the most important consideration, the patch can't land without
>>>> this being clear.
>>>
>>> This work is being performed under a contract with Xilinx/AMD as
>>> confirmed in a different response to
>>> Simon and therefore covered by the copyright assignment.
>>>
>>> I will make the changes as described below and retest.
>>>
>>> Approved for commit ?
>>>
>>> dhanyavaadaaha
>>> gopi
>>>
>>>>
>>>>>   From a3328b4cdfc58783b5f290f22255b36e26a6fb14 Mon Sep 17 00:00:00 2001
>>>>> From: Gopi Kumar Bulusu <gopi@sankhya.com>
>>>>> Date: Thu, 7 May 2026 17:47:40 +0530
>>>>> Subject: [PATCH] gdb/MicroBlaze: Add support for native linux gdb
>>>>
>>>> It's best to just git send-email rather than using an attachment.
>>>>
>>>>>       bfd/ChangeLog:
>>>>
>>>>>       * elf32-microblaze.c (microblaze_elf_grok_prstatus): New function.
>>>>>         (microblaze_elf_grok_psinfo): Likewise.
>>>>
>>>>>       gdb/ChangeLog:
>>>>
>>>>>       * Makefile.in (HFILES_NO_SRCDIR): Add microblaze-linux-tdep.h
>>>>>         ALLDEPFILES: Add microblaze-linux-nat.c
>>>>
>>>> gdb doesn't use ChangeLogs any more and this text should all be removed.
>>>>
>>>>> +#define MICROBLAZE_TARGET_HAS_GETREGS 0
>>>>> +#define MICROBLAZE_TARGET_HAS_SETREGS 0
>>>>
>>>> This seems weird:
>>>>
>>>>> +/* Wrapper function around ptrace.  */
>>>>> +
>>>>> +static void
>>>>> +fetch_target_gp_regs (int tid, elf_gregset_t *gregs)
>>>>> +{
>>>>> +  elf_greg_t *gregp = *gregs;
>>>>> +
>>>>> +#if MICROBLAZE_TARGET_HAS_GETREGS
>>>>> +  if (ptrace (PTRACE_GETREGS, tid, 0, (long) gregp) < 0)
>>>>> +    {
>>>>> +      perror_with_name (_("Couldn't get registers"));
>>>>> +      return;
>>>>> +    }
>>>>> +#error "this configuration did not work during testing"
>>>>
>>>> Normally we don't put new dead code into gdb.
>>>>
>>>> I think this and the defines could just be replaced with a comment
>>>> explaining that PTRACE_GETREGS doesn't work on this platform.
>>>>
>>>>> +++ b/gdb/microblaze-linux-tdep.h
>>>>> @@ -0,0 +1,24 @@
>>>>> +/* Target-dependent code for GNU/Linux on MicroBlaze.
>>>>> +
>>>>> +   Copyright (C) 2021-2026 Free Software Foundation, Inc.
>>>>> +
>>>>> +   This file is part of GDB.
>>>>> +
>>>>> +   This program is free software; you can redistribute it and/or modify
>>>>> +   it under the terms of the GNU General Public License as published by
>>>>> +   the Free Software Foundation; either version 3 of the License, or
>>>>> +   (at your option) any later version.
>>>>> +
>>>>> +   This program is distributed in the hope that it will be useful,
>>>>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>>>>> +   GNU General Public License for more details.
>>>>> +
>>>>> +   You should have received a copy of the GNU General Public License
>>>>> +   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
>>>>> +#ifndef MICROBLAZE_LINUX_TDEP_H
>>>>> +#define MICROBLAZE_LINUX_TDEP_H
>>>>> + /* Target descriptions.  */
>>>>> + extern struct target_desc *tdesc_microblaze_linux;
>>>>
>>>> Blank line between the #define and the comment.
>>>>
>>>> Also this define should have a slightly different name, see
>>>> check-include-guards.py.
>>>>
>>>>>           non_stack_instruction_found = 0;
>>>>> +          cache->register_offsets[rd] = -imm;
>>>>>           continue;
>>>>
>>>> Indentation looks wrong, probably not using tabs.
>>>>
>>>>> @@ -399,8 +400,7 @@ microblaze_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR start_pc)
>>>>>        {
>>>>>          sal = find_sal_for_pc (func_start, 0);
>>>>
>>>>> -      if (sal.end < func_end
>>>>> -       && start_pc <= sal.end)
>>>>> +      if (sal.line != 0 && sal.end <= func_end  && start_pc <= sal.end)
>>>>>         start_pc = sal.end;
>>>>
>>>> No objection from me but this seems somewhat unrelated.
>>>>
>>>>
>>>>>      /* Call for side effects.  */
>>>>> -  get_frame_func (next_frame);
>>>>> +  cache->pc = get_frame_func (next_frame);
>>>>
>>>> Please remove that comment as it is now wrong.
>>>>
>>>>> +  if (regnum == MICROBLAZE_SP_REGNUM)
>>>>> +        regnum = 1;
>>>>
>>>> Wrong indentation.
>>>>
>>>>> +  if (regnum == -1) {
>>>>> +    int i;
>>>>> +
>>>>> +    for (i = 0; i < MICROBLAZE_REDR_REGNUM; i++) {
>>>>> +      regcache->raw_supply (i, regs + i * MICROBLAZE_REGISTER_SIZE);
>>>>> +    }
>>>>
>>>> Wrong brace placement in this hunk.
>>>>
>>>>> +  cb (".reg", tdep->sizeof_gregset, tdep->sizeof_gregset, tdep->gregset, NULL,
>>>>> +      cb_data);
>>>>
>>>> Normally in new code we're using 'nullptr' now.  Best would be to go
>>>> through the whole patch and check for this.  You don't have to fix
>>>> pre-existing code that is using NULL though.
>>>>
>>>>>      gdbarch *gdbarch
>>>>>        = gdbarch_alloc (&info, gdbarch_tdep_up (new microblaze_gdbarch_tdep));
>>>>
>>>>> +  microblaze_gdbarch_tdep *tdep
>>>>> +    = gdbarch_tdep<microblaze_gdbarch_tdep> (gdbarch);
>>>>> +
>>>>> +  tdep->gregset = NULL;
>>>>> +  tdep->sizeof_gregset = 0;
>>>>> +  tdep->fpregset = NULL;
>>>>> +  tdep->sizeof_fpregset = 0;
>>>>
>>>> Here it would be better to just add inline initializers to the new
>>>> fields in microblaze_gdbarch_tdep.  Then they'll automatically be
>>>> initialized by the 'new'.
>>>>
>>>> Tom
>>>
>>
  

namaskaaram

On Thu, Jun 11, 2026, 6:09 AM Michael Eager <eager@eagercon.com> wrote:

> Hi Gopi --
>
> I (or more accurately, git) fixed several whitespace issues.
>

Please share. I did check all the whitespace issues flagged by git-diff.


> Since this is a revised patch, has it been retested?
>


Yes, it has been retested.

dhanyavaadaaha
gopi


> On 6/1/26 10:05 PM, Gopi Kumar Bulusu wrote:
> > namaskaaram
> >
> > Attached is the updated patch. All the changes suggested on PATCH v1
> > have been made and tested.
> >
> > I did not remove the (sal.line != 0) check as it seemed trivial.
> > However, I can make that a separate patch if necessary.
> >
> > Our sysadm In the process of setting up a git compatible email - the
> > process may take more time; until then appreciate your patience with
> > the attachments.
> >
> > Important Note:
> >
> > 1) gdb appears to have an elaborate scheme to handle C++ exceptions
> > and "tunnel" them through exception unaware C code. This code is with
> > readline. Anytime
> > an error is discovered (even a syntax error) by a module,
> > gdb_exception_error() thrown is tunneled through readline using the
> > TRY_SJLJ/CATCH_SJLJ macros
> > implemented using setjmp/longjmp.
> >
> > 2) There is an issue that I came across while testing; native gdb goes
> > into an infinite loop when handling errors. I analyzed it and found it
> > to be an obvious glibc
> > sysdeps/unix/sysv/linux/microblaze/____longjmp_chk.S issue and
> > reported it to AMD. It is possible I missed out a glibc patch; At this
> > time - this is not a gdb issue.
> >
> > 3) Plan to follow through [2] and ensure a gcc/glibc defect report is
> > created if needed.
> >
> > dhanyavaadaaha
> > gopi
> >
> > On Tue, Jun 2, 2026 at 5:58 AM Michael Eager <eager@eagerm.com> wrote:
> >>
> >> Please post an updated patch for review.
> >>
> >> On 5/31/26 11:38 PM, Gopi Kumar Bulusu wrote:
> >>> On Fri, May 29, 2026 at 8:26 PM Tom Tromey <tom@tromey.com> wrote:
> >>>>
> >>>>>>>>> Gopi Kumar Bulusu <gopi@sankhya.com> writes:
> >>>>
> >>>>> Attached is a patch to add native linux support for gdb including
> >>>>> cache target support..
> >>>>
> >>>> Thanks for the patch.
> >>>>
> >>>> I read through it and have some nits.  There's nothing very serious
> >>>> except this:
> >>>>
> >>>>> The original sources for the patch come from Xilinx/AMD Yocto git
> >>>>> repository (2025.2)
> >>>>
> >>>> Who wrote them and are they covered by the copyright assignment?
> >>>>
> >>>> This is the most important consideration, the patch can't land without
> >>>> this being clear.
> >>>
> >>> This work is being performed under a contract with Xilinx/AMD as
> >>> confirmed in a different response to
> >>> Simon and therefore covered by the copyright assignment.
> >>>
> >>> I will make the changes as described below and retest.
> >>>
> >>> Approved for commit ?
> >>>
> >>> dhanyavaadaaha
> >>> gopi
> >>>
> >>>>
> >>>>>   From a3328b4cdfc58783b5f290f22255b36e26a6fb14 Mon Sep 17 00:00:00
> 2001
> >>>>> From: Gopi Kumar Bulusu <gopi@sankhya.com>
> >>>>> Date: Thu, 7 May 2026 17:47:40 +0530
> >>>>> Subject: [PATCH] gdb/MicroBlaze: Add support for native linux gdb
> >>>>
> >>>> It's best to just git send-email rather than using an attachment.
> >>>>
> >>>>>       bfd/ChangeLog:
> >>>>
> >>>>>       * elf32-microblaze.c (microblaze_elf_grok_prstatus): New
> function.
> >>>>>         (microblaze_elf_grok_psinfo): Likewise.
> >>>>
> >>>>>       gdb/ChangeLog:
> >>>>
> >>>>>       * Makefile.in (HFILES_NO_SRCDIR): Add microblaze-linux-tdep.h
> >>>>>         ALLDEPFILES: Add microblaze-linux-nat.c
> >>>>
> >>>> gdb doesn't use ChangeLogs any more and this text should all be
> removed.
> >>>>
> >>>>> +#define MICROBLAZE_TARGET_HAS_GETREGS 0
> >>>>> +#define MICROBLAZE_TARGET_HAS_SETREGS 0
> >>>>
> >>>> This seems weird:
> >>>>
> >>>>> +/* Wrapper function around ptrace.  */
> >>>>> +
> >>>>> +static void
> >>>>> +fetch_target_gp_regs (int tid, elf_gregset_t *gregs)
> >>>>> +{
> >>>>> +  elf_greg_t *gregp = *gregs;
> >>>>> +
> >>>>> +#if MICROBLAZE_TARGET_HAS_GETREGS
> >>>>> +  if (ptrace (PTRACE_GETREGS, tid, 0, (long) gregp) < 0)
> >>>>> +    {
> >>>>> +      perror_with_name (_("Couldn't get registers"));
> >>>>> +      return;
> >>>>> +    }
> >>>>> +#error "this configuration did not work during testing"
> >>>>
> >>>> Normally we don't put new dead code into gdb.
> >>>>
> >>>> I think this and the defines could just be replaced with a comment
> >>>> explaining that PTRACE_GETREGS doesn't work on this platform.
> >>>>
> >>>>> +++ b/gdb/microblaze-linux-tdep.h
> >>>>> @@ -0,0 +1,24 @@
> >>>>> +/* Target-dependent code for GNU/Linux on MicroBlaze.
> >>>>> +
> >>>>> +   Copyright (C) 2021-2026 Free Software Foundation, Inc.
> >>>>> +
> >>>>> +   This file is part of GDB.
> >>>>> +
> >>>>> +   This program is free software; you can redistribute it and/or
> modify
> >>>>> +   it under the terms of the GNU General Public License as
> published by
> >>>>> +   the Free Software Foundation; either version 3 of the License, or
> >>>>> +   (at your option) any later version.
> >>>>> +
> >>>>> +   This program is distributed in the hope that it will be useful,
> >>>>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> >>>>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> >>>>> +   GNU General Public License for more details.
> >>>>> +
> >>>>> +   You should have received a copy of the GNU General Public License
> >>>>> +   along with this program.  If not, see <
> http://www.gnu.org/licenses/>.  */
> >>>>> +#ifndef MICROBLAZE_LINUX_TDEP_H
> >>>>> +#define MICROBLAZE_LINUX_TDEP_H
> >>>>> + /* Target descriptions.  */
> >>>>> + extern struct target_desc *tdesc_microblaze_linux;
> >>>>
> >>>> Blank line between the #define and the comment.
> >>>>
> >>>> Also this define should have a slightly different name, see
> >>>> check-include-guards.py.
> >>>>
> >>>>>           non_stack_instruction_found = 0;
> >>>>> +          cache->register_offsets[rd] = -imm;
> >>>>>           continue;
> >>>>
> >>>> Indentation looks wrong, probably not using tabs.
> >>>>
> >>>>> @@ -399,8 +400,7 @@ microblaze_skip_prologue (struct gdbarch
> *gdbarch, CORE_ADDR start_pc)
> >>>>>        {
> >>>>>          sal = find_sal_for_pc (func_start, 0);
> >>>>
> >>>>> -      if (sal.end < func_end
> >>>>> -       && start_pc <= sal.end)
> >>>>> +      if (sal.line != 0 && sal.end <= func_end  && start_pc <=
> sal.end)
> >>>>>         start_pc = sal.end;
> >>>>
> >>>> No objection from me but this seems somewhat unrelated.
> >>>>
> >>>>
> >>>>>      /* Call for side effects.  */
> >>>>> -  get_frame_func (next_frame);
> >>>>> +  cache->pc = get_frame_func (next_frame);
> >>>>
> >>>> Please remove that comment as it is now wrong.
> >>>>
> >>>>> +  if (regnum == MICROBLAZE_SP_REGNUM)
> >>>>> +        regnum = 1;
> >>>>
> >>>> Wrong indentation.
> >>>>
> >>>>> +  if (regnum == -1) {
> >>>>> +    int i;
> >>>>> +
> >>>>> +    for (i = 0; i < MICROBLAZE_REDR_REGNUM; i++) {
> >>>>> +      regcache->raw_supply (i, regs + i * MICROBLAZE_REGISTER_SIZE);
> >>>>> +    }
> >>>>
> >>>> Wrong brace placement in this hunk.
> >>>>
> >>>>> +  cb (".reg", tdep->sizeof_gregset, tdep->sizeof_gregset,
> tdep->gregset, NULL,
> >>>>> +      cb_data);
> >>>>
> >>>> Normally in new code we're using 'nullptr' now.  Best would be to go
> >>>> through the whole patch and check for this.  You don't have to fix
> >>>> pre-existing code that is using NULL though.
> >>>>
> >>>>>      gdbarch *gdbarch
> >>>>>        = gdbarch_alloc (&info, gdbarch_tdep_up (new
> microblaze_gdbarch_tdep));
> >>>>
> >>>>> +  microblaze_gdbarch_tdep *tdep
> >>>>> +    = gdbarch_tdep<microblaze_gdbarch_tdep> (gdbarch);
> >>>>> +
> >>>>> +  tdep->gregset = NULL;
> >>>>> +  tdep->sizeof_gregset = 0;
> >>>>> +  tdep->fpregset = NULL;
> >>>>> +  tdep->sizeof_fpregset = 0;
> >>>>
> >>>> Here it would be better to just add inline initializers to the new
> >>>> fields in microblaze_gdbarch_tdep.  Then they'll automatically be
> >>>> initialized by the 'new'.
> >>>>
> >>>> Tom
> >>>
> >>
>
> --
> Michael Eager
>
>
  

On 6/10/26 9:19 PM, Gopi Kumar Bulusu wrote:
> 
> namaskaaram 
> 
> On Thu, Jun 11, 2026, 6:09 AM Michael Eager <eager@eagercon.com <mailto:eager@eagercon.com>> wrote:
> 
>     Hi Gopi --
> 
>     I (or more accurately, git) fixed several whitespace issues.
> 
> 
> Please share. I did check all the whitespace issues flagged by git-diff.

For whitespaces, I suggest to install pre-commit, there is a hook that
checks it.  See:

https://sourceware.org/gdb/wiki/DeveloperTips#Setting_up_pre-commit

Simon
  

namaskaaram

On Thu, Jun 11, 2026, 7:11 AM Simon Marchi <simark@simark.ca> wrote:

> On 6/10/26 9:19 PM, Gopi Kumar Bulusu wrote:
> >
> > namaskaaram
> >
> > On Thu, Jun 11, 2026, 6:09 AM Michael Eager <eager@eagercon.com <mailto:
> eager@eagercon.com>> wrote:
> >
> >     Hi Gopi --
> >
> >     I (or more accurately, git) fixed several whitespace issues.
> >
> >
> > Please share. I did check all the whitespace issues flagged by git-diff.
>
> For whitespaces, I suggest to install pre-commit, there is a hook that
> checks it.  See:
>
> https://sourceware.org/gdb/wiki/DeveloperTips#Setting_up_pre-commit


Thanks, I will install  it.

dhanyavaadaaha
gopi


> Simon
>
  

Committed d62f64e17

On 6/10/26 6:19 PM, Gopi Kumar Bulusu wrote:
> 
> namaskaaram
> 
> On Thu, Jun 11, 2026, 6:09 AM Michael Eager <eager@eagercon.com 
> <mailto:eager@eagercon.com>> wrote:
> 
>     Hi Gopi --
> 
>     I (or more accurately, git) fixed several whitespace issues.
> 
> 
> Please share. I did check all the whitespace issues flagged by git-diff.
> 
> 
>     Since this is a revised patch, has it been retested?
> 
> 
> 
> Yes, it has been retested.
> 
> dhanyavaadaaha
> gopi
> 
> 
>     On 6/1/26 10:05 PM, Gopi Kumar Bulusu wrote:
>      > namaskaaram
>      >
>      > Attached is the updated patch. All the changes suggested on PATCH v1
>      > have been made and tested.
>      >
>      > I did not remove the (sal.line != 0) check as it seemed trivial.
>      > However, I can make that a separate patch if necessary.
>      >
>      > Our sysadm In the process of setting up a git compatible email - the
>      > process may take more time; until then appreciate your patience with
>      > the attachments.
>      >
>      > Important Note:
>      >
>      > 1) gdb appears to have an elaborate scheme to handle C++ exceptions
>      > and "tunnel" them through exception unaware C code. This code is with
>      > readline. Anytime
>      > an error is discovered (even a syntax error) by a module,
>      > gdb_exception_error() thrown is tunneled through readline using the
>      > TRY_SJLJ/CATCH_SJLJ macros
>      > implemented using setjmp/longjmp.
>      >
>      > 2) There is an issue that I came across while testing; native gdb
>     goes
>      > into an infinite loop when handling errors. I analyzed it and
>     found it
>      > to be an obvious glibc
>      > sysdeps/unix/sysv/linux/microblaze/____longjmp_chk.S issue and
>      > reported it to AMD. It is possible I missed out a glibc patch; At
>     this
>      > time - this is not a gdb issue.
>      >
>      > 3) Plan to follow through [2] and ensure a gcc/glibc defect report is
>      > created if needed.
>      >
>      > dhanyavaadaaha
>      > gopi
>      >
>      > On Tue, Jun 2, 2026 at 5:58 AM Michael Eager <eager@eagerm.com
>     <mailto:eager@eagerm.com>> wrote:
>      >>
>      >> Please post an updated patch for review.
>      >>
>      >> On 5/31/26 11:38 PM, Gopi Kumar Bulusu wrote:
>      >>> On Fri, May 29, 2026 at 8:26 PM Tom Tromey <tom@tromey.com
>     <mailto:tom@tromey.com>> wrote:
>      >>>>
>      >>>>>>>>> Gopi Kumar Bulusu <gopi@sankhya.com
>     <mailto:gopi@sankhya.com>> writes:
>      >>>>
>      >>>>> Attached is a patch to add native linux support for gdb including
>      >>>>> cache target support..
>      >>>>
>      >>>> Thanks for the patch.
>      >>>>
>      >>>> I read through it and have some nits.  There's nothing very
>     serious
>      >>>> except this:
>      >>>>
>      >>>>> The original sources for the patch come from Xilinx/AMD Yocto git
>      >>>>> repository (2025.2)
>      >>>>
>      >>>> Who wrote them and are they covered by the copyright assignment?
>      >>>>
>      >>>> This is the most important consideration, the patch can't land
>     without
>      >>>> this being clear.
>      >>>
>      >>> This work is being performed under a contract with Xilinx/AMD as
>      >>> confirmed in a different response to
>      >>> Simon and therefore covered by the copyright assignment.
>      >>>
>      >>> I will make the changes as described below and retest.
>      >>>
>      >>> Approved for commit ?
>      >>>
>      >>> dhanyavaadaaha
>      >>> gopi
>      >>>
>      >>>>
>      >>>>>   From a3328b4cdfc58783b5f290f22255b36e26a6fb14 Mon Sep 17
>     00:00:00 2001
>      >>>>> From: Gopi Kumar Bulusu <gopi@sankhya.com
>     <mailto:gopi@sankhya.com>>
>      >>>>> Date: Thu, 7 May 2026 17:47:40 +0530
>      >>>>> Subject: [PATCH] gdb/MicroBlaze: Add support for native linux gdb
>      >>>>
>      >>>> It's best to just git send-email rather than using an attachment.
>      >>>>
>      >>>>>       bfd/ChangeLog:
>      >>>>
>      >>>>>       * elf32-microblaze.c (microblaze_elf_grok_prstatus):
>     New function.
>      >>>>>         (microblaze_elf_grok_psinfo): Likewise.
>      >>>>
>      >>>>>       gdb/ChangeLog:
>      >>>>
>      >>>>>       * Makefile.in (HFILES_NO_SRCDIR): Add microblaze-linux-
>     tdep.h
>      >>>>>         ALLDEPFILES: Add microblaze-linux-nat.c
>      >>>>
>      >>>> gdb doesn't use ChangeLogs any more and this text should all
>     be removed.
>      >>>>
>      >>>>> +#define MICROBLAZE_TARGET_HAS_GETREGS 0
>      >>>>> +#define MICROBLAZE_TARGET_HAS_SETREGS 0
>      >>>>
>      >>>> This seems weird:
>      >>>>
>      >>>>> +/* Wrapper function around ptrace.  */
>      >>>>> +
>      >>>>> +static void
>      >>>>> +fetch_target_gp_regs (int tid, elf_gregset_t *gregs)
>      >>>>> +{
>      >>>>> +  elf_greg_t *gregp = *gregs;
>      >>>>> +
>      >>>>> +#if MICROBLAZE_TARGET_HAS_GETREGS
>      >>>>> +  if (ptrace (PTRACE_GETREGS, tid, 0, (long) gregp) < 0)
>      >>>>> +    {
>      >>>>> +      perror_with_name (_("Couldn't get registers"));
>      >>>>> +      return;
>      >>>>> +    }
>      >>>>> +#error "this configuration did not work during testing"
>      >>>>
>      >>>> Normally we don't put new dead code into gdb.
>      >>>>
>      >>>> I think this and the defines could just be replaced with a comment
>      >>>> explaining that PTRACE_GETREGS doesn't work on this platform.
>      >>>>
>      >>>>> +++ b/gdb/microblaze-linux-tdep.h
>      >>>>> @@ -0,0 +1,24 @@
>      >>>>> +/* Target-dependent code for GNU/Linux on MicroBlaze.
>      >>>>> +
>      >>>>> +   Copyright (C) 2021-2026 Free Software Foundation, Inc.
>      >>>>> +
>      >>>>> +   This file is part of GDB.
>      >>>>> +
>      >>>>> +   This program is free software; you can redistribute it
>     and/or modify
>      >>>>> +   it under the terms of the GNU General Public License as
>     published by
>      >>>>> +   the Free Software Foundation; either version 3 of the
>     License, or
>      >>>>> +   (at your option) any later version.
>      >>>>> +
>      >>>>> +   This program is distributed in the hope that it will be
>     useful,
>      >>>>> +   but WITHOUT ANY WARRANTY; without even the implied
>     warranty of
>      >>>>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>      >>>>> +   GNU General Public License for more details.
>      >>>>> +
>      >>>>> +   You should have received a copy of the GNU General Public
>     License
>      >>>>> +   along with this program.  If not, see <http://
>     www.gnu.org/licenses/ <http://www.gnu.org/licenses/>>.  */
>      >>>>> +#ifndef MICROBLAZE_LINUX_TDEP_H
>      >>>>> +#define MICROBLAZE_LINUX_TDEP_H
>      >>>>> + /* Target descriptions.  */
>      >>>>> + extern struct target_desc *tdesc_microblaze_linux;
>      >>>>
>      >>>> Blank line between the #define and the comment.
>      >>>>
>      >>>> Also this define should have a slightly different name, see
>      >>>> check-include-guards.py.
>      >>>>
>      >>>>>           non_stack_instruction_found = 0;
>      >>>>> +          cache->register_offsets[rd] = -imm;
>      >>>>>           continue;
>      >>>>
>      >>>> Indentation looks wrong, probably not using tabs.
>      >>>>
>      >>>>> @@ -399,8 +400,7 @@ microblaze_skip_prologue (struct gdbarch
>     *gdbarch, CORE_ADDR start_pc)
>      >>>>>        {
>      >>>>>          sal = find_sal_for_pc (func_start, 0);
>      >>>>
>      >>>>> -      if (sal.end < func_end
>      >>>>> -       && start_pc <= sal.end)
>      >>>>> +      if (sal.line != 0 && sal.end <= func_end  && start_pc
>     <= sal.end)
>      >>>>>         start_pc = sal.end;
>      >>>>
>      >>>> No objection from me but this seems somewhat unrelated.
>      >>>>
>      >>>>
>      >>>>>      /* Call for side effects.  */
>      >>>>> -  get_frame_func (next_frame);
>      >>>>> +  cache->pc = get_frame_func (next_frame);
>      >>>>
>      >>>> Please remove that comment as it is now wrong.
>      >>>>
>      >>>>> +  if (regnum == MICROBLAZE_SP_REGNUM)
>      >>>>> +        regnum = 1;
>      >>>>
>      >>>> Wrong indentation.
>      >>>>
>      >>>>> +  if (regnum == -1) {
>      >>>>> +    int i;
>      >>>>> +
>      >>>>> +    for (i = 0; i < MICROBLAZE_REDR_REGNUM; i++) {
>      >>>>> +      regcache->raw_supply (i, regs + i *
>     MICROBLAZE_REGISTER_SIZE);
>      >>>>> +    }
>      >>>>
>      >>>> Wrong brace placement in this hunk.
>      >>>>
>      >>>>> +  cb (".reg", tdep->sizeof_gregset, tdep->sizeof_gregset,
>     tdep->gregset, NULL,
>      >>>>> +      cb_data);
>      >>>>
>      >>>> Normally in new code we're using 'nullptr' now.  Best would be
>     to go
>      >>>> through the whole patch and check for this.  You don't have to fix
>      >>>> pre-existing code that is using NULL though.
>      >>>>
>      >>>>>      gdbarch *gdbarch
>      >>>>>        = gdbarch_alloc (&info, gdbarch_tdep_up (new
>     microblaze_gdbarch_tdep));
>      >>>>
>      >>>>> +  microblaze_gdbarch_tdep *tdep
>      >>>>> +    = gdbarch_tdep<microblaze_gdbarch_tdep> (gdbarch);
>      >>>>> +
>      >>>>> +  tdep->gregset = NULL;
>      >>>>> +  tdep->sizeof_gregset = 0;
>      >>>>> +  tdep->fpregset = NULL;
>      >>>>> +  tdep->sizeof_fpregset = 0;
>      >>>>
>      >>>> Here it would be better to just add inline initializers to the new
>      >>>> fields in microblaze_gdbarch_tdep.  Then they'll automatically be
>      >>>> initialized by the 'new'.
>      >>>>
>      >>>> Tom
>      >>>
>      >>
> 
>     -- 
>     Michael Eager
>