[Bug breakpoints/26619] New: [arm] Unexpected error setting breakpoint/watchpoint

[Bug breakpoints/26619] New: [arm] Unexpected error setting breakpoint/watchpoint

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

            Bug ID: 26619
           Summary: [arm] Unexpected error setting breakpoint/watchpoint
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: breakpoints
          Assignee: unassigned at sourceware dot org
          Reporter: vries at gcc dot gnu.org
  Target Milestone: ---

On armv7hl, I get:
...
$ egrep -ca "Unexpected error setting (breakpoint|watchpoint): No such device"
binaries-testsuite.*/usr/share/doc/packages/gdb-testresults/*.log | grep -v :0
binaries-testsuite.openSUSE_Factory_ARM.armv7l/usr/share/doc/packages/gdb-testresults/gdb-armv7hl-suse-linux.-fno-PIE.-no-pie.log:582
binaries-testsuite.openSUSE_Factory_ARM.armv7l/usr/share/doc/packages/gdb-testresults/gdb-armv7hl-suse-linux.log:583
binaries-testsuite.openSUSE_Leap_15.1.armv7l/usr/share/doc/packages/gdb-testresults/gdb-armv7hl-suse-linux.-fno-PIE.-no-pie.log:579
binaries-testsuite.openSUSE_Leap_15.1.armv7l/usr/share/doc/packages/gdb-testresults/gdb-armv7hl-suse-linux.log:585
binaries-testsuite.openSUSE_Leap_15.2.armv7l/usr/share/doc/packages/gdb-testresults/gdb-armv7hl-suse-linux.-fno-PIE.-no-pie.log:579
binaries-testsuite.openSUSE_Leap_15.2.armv7l/usr/share/doc/packages/gdb-testresults/gdb-armv7hl-suse-linux.log:588
...

F.i., we have:
...
(gdb) PASS: gdb.base/break-idempotent.exp: pie=nopie: always_inserted=on:
hbreak: b bar
continue^M
Continuing.^M
Unexpected error setting breakpoint: No such device.^M
(gdb) FAIL: gdb.base/break-idempotent.exp: pie=nopie: always_inserted=on:
hbreak: continue
...

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |alahay01 at gcc dot gnu.org,
                   |                            |luis.machado at linaro dot org

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[luis.machado at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

--- Comment #1 from Luis Machado <luis.machado at linaro dot org> ---
I'm aware of various hardware watchpoint issues for arm 32-bit. I think this is
one of the manifestations. It depends on the kernel version as well, since
there has been fixes to the kernel.

What exact hardware is this by the way?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

--- Comment #2 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Luis Machado from comment #1)
> I'm aware of various hardware watchpoint issues for arm 32-bit. I think this
> is one of the manifestations. It depends on the kernel version as well,
> since there has been fixes to the kernel.
> 

I see.  It also happens on Factory though, where I'd expect a reasonably
up-2-date kernel.

> What exact hardware is this by the way?

I'm not sure, it's OBS and AFAICT from the log it's KVM/QEMU.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

Thiago Jung Bauermann <thiago.bauermann at linaro dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |thiago.bauermann at linaro dot org

--- Comment #3 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> ---
I just came across this bug report and grepped the random collection of
testsuite log files have accumulated on my computer.

I only see this "No such device" message on emulated QEMU arm VMs, not on real
hardware (which is armv8l userspace running on 64-bit kernel on Neoverse-N1).
And even then, it's not all results on emulated QEMU arm, just some of them. I
don't think I have any result from KVM.

So as a first approximation, this appears related to QEMU in some way.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[luis.machado at arm dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

--- Comment #4 from Luis Machado <luis.machado at arm dot com> ---
I think this used to be the case on older kernels. One of the fixes was to
actually try to insert the hardware watchpoint (say, on PC), and see if that
succeeds. If so, then go ahead and use it. Otherwise, assume it doesn't work.

If this isn't manifesting anymore, maybe we should just close it.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

--- Comment #5 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> ---
For the record, the testsuite results where I found this error are from
September 2022, with kernel:

Linux debian-arm 5.10.0-17-armmp-lpae #1 SMP Debian 5.10.136-1 (2022-08-13)
armv7l GNU/Linux

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

--- Comment #6 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> ---
The testcases which ran into that "Unexpected error setting breakpoint: No
such device." in my gdb.log file where:

- gdb.base/break-idempotent.exp
- gdb.base/break-unload-file.exp
- gdb.base/breakpoint-in-ro-region.exp
- gdb.base/hbreak.exp
- gdb.base/hbreak2.exp
- gdb.python/py-breakpoint.exp
- gdb.threads/watchpoint-fork.exp

I updated that QEMU arm TCG VM to Debian 12 (bookworm) and installed the
newest kernel from Debian unstable, which is 6.7.9-armmp-lpae.

Now none of the tests above have that failure. But gdb.base/hbreak.exp has
a different failure now:

Continuing.
Unexpected error setting breakpoint: Invalid argument.
(gdb) FAIL: gdb.base/hbreak.exp: continue to break-at-exit after hbreak

I also see this failure on real hardware with armv8l-linux and Ubuntu's
5.4.0-131-generic kernel. This looks like a different issue.

All the other tests are passing. I'll leave them running in a loop
overnight to confirm (since not all log files from the old testsuite runs
on that machine show this problem).

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

--- Comment #7 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> ---
(In reply to Thiago Jung Bauermann from comment #6)
> All the other tests are passing. I'll leave them running in a loop
> overnight to confirm (since not all log files from the old testsuite runs
> on that machine show this problem).

I had no test failures overnight, so I can officially say that I can't
reproduce the problem anymore.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |UNCONFIRMED
     Ever confirmed|1                           |0

--- Comment #8 from Tom de Vries <vries at gcc dot gnu.org> ---
Thanks for checking this Thiago.

Then I propose to mark this resolved-worksforme.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/26619] [arm] Unexpected error setting breakpoint/watchpoint

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=26619

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |WORKSFORME

--- Comment #9 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #8)
> Thanks for checking this Thiago.
> 
> Then I propose to mark this resolved-worksforme.

Like so.

-- 
You are receiving this mail because:
You are on the CC list for the bug.