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 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 wrote: > >> > >>>>>>> Gopi Kumar Bulusu 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 > >>> 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 . */ > >>> +#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 (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 > > >