[Bug gdb/23346] New: [aarch64] gdb assertion: inf != NULL

[wdijkstr at arm dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23346

Bug ID: 23346
Summary: [aarch64] gdb assertion: inf != NULL
Product: gdb
Version: HEAD
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: gdb
Assignee: unassigned at sourceware dot org
Reporter: wdijkstr at arm dot com
Target Milestone: ---

The following tests fail on recent gdb:

FAIL: gdb.base/foll-exec-mode.exp: new,continue,infswitch: continue past exec
(GDB internal error)
FAIL: gdb.base/foll-exec-mode.exp: new,continue,no_infswitch: continue past
exec (GDB internal error)
FAIL: gdb.base/foll-exec-mode.exp: new,next,infswitch: next past exec (GDB
internal error)
FAIL: gdb.base/watchpoint-hw-attach.exp: continue (the program is no longer
running)
FAIL: gdb.base/watchpoint-hw-attach.exp: continue to breakpoint:
prewatchtrigger (the program exited)
FAIL: gdb.base/watchpoint-hw-attach.exp: setting breakpoint at 39

These trigger an assertion:
next

process 12644 is executing new program:
buildbot/native-glibc-aarch64-none-linux-gnu/build/build-native-aarch64-none-linux-gnu/obj/gdb/gdb/testsuite/outputs/gdb.base/foll-exec-mode/execd-prog

[New inferior 2 (process 0)]

[New process 12644]

buildbot/native-glibc-aarch64-none-linux-gnu/build/src/binutils-gdb--gdb/gdb/target.c:3200:
internal-error: gdbarch* default_thread_architecture(target_ops*, ptid_t):
Assertion `inf != NULL' failed.

A problem internal to GDB has been detected,


This appears to be due to a recent commit which changes HW breakpoint
attachment.

[wdijkstr at arm dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23346

Wilco <wdijkstr at arm dot com> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |richard.bunt at arm dot com

[richard.bunt at arm dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23346

--- Comment #1 from Richard Bunt <richard.bunt at arm dot com> ---
(In reply to Wilco from comment #0)

Post by wdijkstr at arm dot com
FAIL: gdb.base/foll-exec-mode.exp: new,continue,infswitch: continue past
exec (GDB internal error)
FAIL: gdb.base/foll-exec-mode.exp: new,continue,no_infswitch: continue past
exec (GDB internal error)
FAIL: gdb.base/foll-exec-mode.exp: new,next,infswitch: next past exec (GDB
internal error)
FAIL: gdb.base/watchpoint-hw-attach.exp: continue (the program is no longer
running)
prewatchtrigger (the program exited)
FAIL: gdb.base/watchpoint-hw-attach.exp: setting breakpoint at 39
next
buildbot/native-glibc-aarch64-none-linux-gnu/build/build-native-aarch64-none-
linux-gnu/obj/gdb/gdb/testsuite/outputs/gdb.base/foll-exec-mode/execd-prog
[New inferior 2 (process 0)]
[New process 12644]
buildbot/native-glibc-aarch64-none-linux-gnu/build/src/binutils-gdb--gdb/gdb/
target.c:3200: internal-error: gdbarch*
default_thread_architecture(target_ops*, ptid_t): Assertion `inf != NULL'
failed.
A problem internal to GDB has been detected,
This appears to be due to a recent commit which changes HW breakpoint
attachment.

Thanks for the report.

watchpoint-hw-attach.exp is failing due to a broken assumption in the test
case. I've got a fix ready to go for this.

I will investigate the cause of the gdb.base/foll-exec-mode.exp failure.

[richard.bunt at arm dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23346

--- Comment #2 from Richard Bunt <richard.bunt at arm dot com> ---
watchpoint-hw-attach.exp:

Fix pushed in be1b6474f0d471c01f5c4d1f33e4f13df9ffb82c

gdb.base/foll-exec-mode.exp:

Performing a bisect:

$ git bisect log
git bisect start
# good: [57c0d77c2ce5e583dab322e05f8291bcbad0ccd3] Use offsets instead of
addresses in ELF_SECTION_IN_SEGMENT for non SHT_NOBITS
git bisect good 57c0d77c2ce5e583dab322e05f8291bcbad0ccd3
# bad: [1021d1cb1214771711c61117ba0923c73a3ff431] Add a syntax table to
dwarf-mode.el
git bisect bad 1021d1cb1214771711c61117ba0923c73a3ff431
# bad: [6c04fd9b2fb4396c0189cb414ce598161ac8673e] Add support for
.note.gnu.property sections.
git bisect bad 6c04fd9b2fb4396c0189cb414ce598161ac8673e
# bad: [334cd8a780840d6aee86790fe2fe5f02fe56edf6] MIPS/BFD: Keep EI_ABIVERSION
setting together
git bisect bad 334cd8a780840d6aee86790fe2fe5f02fe56edf6
# good: [33bab475a6984afedac8a036b3bb40b5555b4127] Avoid memcpys in regcache
read_part/write_part for full registers.
git bisect good 33bab475a6984afedac8a036b3bb40b5555b4127
# bad: [25eb8346acc04b10d62295eb4c67c928b01a8aaf] RISC-V: Fix 3 PIE related ld
testsuite failures.
git bisect bad 25eb8346acc04b10d62295eb4c67c928b01a8aaf
# bad: [00431a78b28f913a9d5c912c49680e39cfd20847] Use thread_info and inferior
pointers more throughout
git bisect bad 00431a78b28f913a9d5c912c49680e39cfd20847
# first bad commit: [00431a78b28f913a9d5c912c49680e39cfd20847] Use thread_info
and inferior pointers more throughout


indicates that 00431a78b28f913a9d5c912c49680e39cfd20847 is likely to be the bad
commit rather than the change to hardware watchpoints on attach.

I've confirmed this regression point with clean builds of
00431a78b28f913a9d5c912c49680e39cfd20847 and
33bab475a6984afedac8a036b3bb40b5555b4127

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23346

Pedro Alves <palves at redhat dot com> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |palves at redhat dot com

--- Comment #3 from Pedro Alves <palves at redhat dot com> ---
Probably the same as:
https://sourceware.org/ml/gdb-patches/2018-06/msg00660.html
(Haven't managed to look at it, busy fixing other regressions still.)

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23346

Pedro Alves <palves at redhat dot com> changed:

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

--- Comment #4 from Pedro Alves <palves at redhat dot com> ---
Should be fixed now:
https://sourceware.org/ml/gdb-patches/2018-06/msg00687.html