[gdb/testsuite] Add xfail in gdb.python/py-record-btrace.exp
Commit Message
[ was: Re: [PATCH] [gdb/testsuite] Add have_linux_btrace_bug ]
On 2/13/23 20:17, Metzger, Markus T wrote:
> Hello Tom,
>
>> The linux kernel commit 670638477aed ("perf/x86/intel/pt: Opportunistically
>> use single range output mode"), added in version v5.5.0 had a bug that was
>> fixed by commit ce0d998be927 ("perf/x86/intel/pt: Fix sampling using
>> single range output") in version 6.1.0.
>>
>> The bug manifested for intel microarchitectures Rocket Lake, Raptor Lake and
>> Alder Lake.
>
> Actually, it's a h/w bug that got exposed by using single-range output. It affects
> Core processors starting from Ice Lake and it only affects Processor Trace. Also,
> it is only exposed by the py-record-btrace test, which does a lot of single-stepping.
>
> It might be better to just add an XFAIL for that one test. I'm not sure if maintaining
> a processor list makes sense. The kernel patch disables single-range for > 1 page
> for all processors and does not try to maintain a list of affected processors. We might
> want to do the same in GDB and either disable that test for kernels between 5.5 and
> 6.1, or setup an XFAIL.
>
I've gone with the XFAIL, see attached patch. WDYT?
> The Branch Trace Store issue you found seems to affect all btrace tests on ADL
> E-cores. This is a different issue. I can reproduce it and I am currently debugging it.
I see, ack, thanks for the update.
- Tom
Comments
Hello Tom,
+ set have_xfail 0
+ set v [linux_kernel_version]
+ if { $v == {} } {
+ set have_xfail 0
It is already initialized to 0 so we may turn this into
If { $v != {} } {
+ } else {
+ set have_xfail \
+ [expr [version_compare [list 5 5 0] <= $v] \
+ && [version_compare $v < [list 6 1 0]]]
+ }
+ set xfail_re \
+ [join \
+ [list \
+ "warning: Decode error \\(.*\\) at instruction $decimal" \
+ "\\(offset = $hex, pc = $hex\\):" \
+ "compressed return without call\\."]]
The exact error symptom may vary. On my system I got 'bad packet context'.
It should suffice to grep for the first part, without the actual error string.
Looks good to me.
Regards,
Markus.
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
From 99ebac02397f19bc16cad6a6fef7828fbc716803 Mon Sep 17 00:00:00 2001
From: Tom de Vries <tdevries@suse.de>
Date: Mon, 13 Feb 2023 18:27:17 +0100
Subject: [PATCH] [gdb/testsuite] Add xfail in gdb.python/py-record-btrace.exp
There's a HW bug affecting Processor Trace on some Intel processors
(Ice Lake to Raptor Lake microarchitectures).
The bug was exposed by linux kernel commit 670638477aed
("perf/x86/intel/pt: Opportunistically use single range output mode"),
added in version v5.5.0, and was worked around by commit ce0d998be927
("perf/x86/intel/pt: Fix sampling using single range output") in version
6.1.0.
The bug manifests (on a Performance-core of an i7-1250U, an Alder Lake cpu) in
a single test-case:
...
(gdb) python insn = r.instruction_history^M
warning: Decode error (-20) at instruction 33 (offset = 0x3d6a, \
pc = 0x400501): compressed return without call.^M
(gdb) FAIL: gdb.python/py-record-btrace.exp: prepare record: \
python insn = r.instruction_history
...
Add a corresponding XFAIL.
Note that the i7-1250U has both Performance-cores and Efficient-cores, and on
an Efficient-Core the test-case runs without any problems, so if the testsuite
run is not pinned to a specific cpu, the test may either PASS or XFAIL.
Tested on x86_64-linux:
- openSUSE Leap 15.4 with linux kernel version 5.14.21
- openSUSE Tumbleweed with linux kernel version 6.1.8
PR testsuite/30075
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30075
---
gdb/testsuite/gdb.python/py-record-btrace.exp | 44 ++++++++++++++++++-
1 file changed, 43 insertions(+), 1 deletion(-)
@@ -54,7 +54,49 @@ with_test_prefix "prepare record" {
gdb_test "python print(r.method)" "btrace"
gdb_test "python print(r.format)" "pt|bts"
gdb_test "stepi 100" ".*"
- gdb_test_no_output "python insn = r.instruction_history"
+
+ # There's a HW bug affecting Processor Trace on some Intel processors.
+ # The bug was exposed by linux kernel commit 670638477aed
+ # ("perf/x86/intel/pt: Opportunistically use single range output mode"),
+ # added in version v5.5.0, and was worked around by commit ce0d998be927
+ # ("perf/x86/intel/pt: Fix sampling using single range output") in version
+ # 6.1.0. Detect the kernel version range for which the problem may
+ # manifest.
+ set have_xfail 0
+ set v [linux_kernel_version]
+ if { $v == {} } {
+ set have_xfail 0
+ } else {
+ set have_xfail \
+ [expr [version_compare [list 5 5 0] <= $v] \
+ && [version_compare $v < [list 6 1 0]]]
+ }
+ set xfail_re \
+ [join \
+ [list \
+ "warning: Decode error \\(.*\\) at instruction $decimal" \
+ "\\(offset = $hex, pc = $hex\\):" \
+ "compressed return without call\\."]]
+
+ set got_xfail 0
+ set cmd "python insn = r.instruction_history"
+ gdb_test_multiple $cmd "" {
+ -re "^[string_to_regexp $cmd]\r\n$::gdb_prompt $" {
+ pass $gdb_test_name
+ }
+ -re -wrap "$xfail_re" {
+ if { $have_xfail } {
+ xfail $gdb_test_name
+ set got_xfail 1
+ } else {
+ fail $gdb_test_name
+ }
+ }
+ }
+ if { $got_xfail } {
+ return
+ }
+
gdb_test_no_output "python call = r.function_call_history"
gdb_test_no_output "python i = insn\[0\]"
gdb_test_no_output "python c = call\[0\]"
base-commit: 37d75d4552d6b4d292ffb4dee92329a449288375
--
2.35.3