Re: [PATCH] Handle MIPS Linux SIGTRAP siginfo.si_code values

[Pedro Alves <palves at redhat dot com>]

On 04/05/2016 05:57 PM, Maciej W. Rozycki wrote:

>  FYI, the kernel change has now made it upstream:
> 
> commit 3b143cca6e1397188f507a6c727f4108861ceb8b
> Author: Maciej W. Rozycki <macro@imgtec.com>
> Date:   Fri Mar 4 01:44:28 2016 +0000
> 
>     MIPS: traps: Correct the SIGTRAP debug ABI in `do_watch' and `do_trap_or_bp'
> 
> and will be present starting from Linux 4.6.0 release.
> 

Thanks Maciej,

Here's the same patch with updated comments.  I've pushed this to both master
and the 7.11 branch.

I've filed https://sourceware.org/bugzilla/show_bug.cgi?id=19958
(Breakpoints/watchpoints broken on MIPS Linux <= 4.5), as necessary
according to our branch backporting procedure.

>From f25c3723b2251e33e725afa52280fde679f3e600 Mon Sep 17 00:00:00 2001
From: Pedro Alves <palves@redhat.com>
Date: Fri, 15 Apr 2016 23:52:00 +0100
Subject: [PATCH] MIPS/Linux: Also recognize TRAP_BRKPT and TRAP_HWBKPT

This makes the MIPS Linux backends recognize TRAP_BRKPT and
TRAP_HWBKPT in siginfo.si_code in addition to SI_KERNEL, since Linux
4.6 now reports the finer-grained si_code values too.

Refs:
 https://sourceware.org/ml/gdb-patches/2016-02/msg00756.html
 https://sourceware.org/ml/gdb-patches/2016-04/msg00090.html

On kernels that report SI_KERNEL (<= 4.5), we'll enter the "ambiguous"
path of save_stop_reason:

	  if (GDB_ARCH_IS_TRAP_BRKPT (siginfo.si_code)
	      && GDB_ARCH_IS_TRAP_HWBKPT (siginfo.si_code))
	    {
	      /* The si_code is ambiguous on this arch -- check debug
		 registers.  */
	      if (!check_stopped_by_watchpoint (lp))
		lp->stop_reason = TARGET_STOPPED_BY_SW_BREAKPOINT;
	    }

while on kernels that report the finer-grained si_code values (>= 4.6),
we'll enter the corresponding branches:

	  else if (GDB_ARCH_IS_TRAP_BRKPT (siginfo.si_code))
	    {
	    }
	  else if (GDB_ARCH_IS_TRAP_HWBKPT (siginfo.si_code))
	    {
	      ...

gdb/ChangeLog:
2016-04-15  Pedro Alves  <palves@redhat.com>

	* nat/linux-ptrace.h [__mips__] (GDB_ARCH_IS_TRAP_BRKPT): Also
	accept TRAP_BRKPT.
	 [__mips__] (GDB_ARCH_IS_TRAP_HWBKPT): Also accept TRAP_HWBKPT.
---
 gdb/nat/linux-ptrace.h | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/gdb/nat/linux-ptrace.h b/gdb/nat/linux-ptrace.h
index b9123c9..0a23bcb 100644
--- a/gdb/nat/linux-ptrace.h
+++ b/gdb/nat/linux-ptrace.h
@@ -140,8 +140,8 @@ struct buffer;
    in SPU code on a Cell/B.E.  However, SI_KERNEL is never seen
    on a SIGTRAP for any other reason.
 
-   The MIPS kernel uses SI_KERNEL for all kernel generated traps.
-   Since:
+   The MIPS kernel up until 4.5 used SI_KERNEL for all kernel
+   generated traps.  Since:
 
      - MIPS doesn't do hardware single-step.
      - We don't need to care about exec SIGTRAPs --- we assume
@@ -152,6 +152,9 @@ struct buffer;
    software breakpoints and hardware watchpoints, which can be done by
    peeking the debug registers.
 
+   Beginning with Linux 4.6, the MIPS port reports proper TRAP_BRKPT and
+   TRAP_HWBKPT codes, so we also match them.
+
    The generic Linux target code should use GDB_ARCH_IS_TRAP_* instead
    of TRAP_* to abstract out these peculiarities.  */
 #if defined __i386__ || defined __x86_64__
@@ -161,8 +164,8 @@ struct buffer;
 # define GDB_ARCH_IS_TRAP_BRKPT(X) ((X) == SI_KERNEL || (X) == TRAP_BRKPT)
 # define GDB_ARCH_IS_TRAP_HWBKPT(X) ((X) == TRAP_HWBKPT)
 #elif defined __mips__
-# define GDB_ARCH_IS_TRAP_BRKPT(X) ((X) == SI_KERNEL)
-# define GDB_ARCH_IS_TRAP_HWBKPT(X) ((X) == SI_KERNEL)
+# define GDB_ARCH_IS_TRAP_BRKPT(X) ((X) == SI_KERNEL || (X) == TRAP_BRKPT)
+# define GDB_ARCH_IS_TRAP_HWBKPT(X) ((X) == SI_KERNEL || (X) == TRAP_HWBKPT)
 #else
 # define GDB_ARCH_IS_TRAP_BRKPT(X) ((X) == TRAP_BRKPT)
 # define GDB_ARCH_IS_TRAP_HWBKPT(X) ((X) == TRAP_HWBKPT)
-- 
2.5.5