procps: ptest: skip flaky pgrep full process name match test
PATCH 1/1 procps: ptest: skip flaky pgrep full process name match test
The 'pgrep match against full process name' test uses pgrep -f with a regex pattern close to ARG_MAX in size (~100KB). This intermittently fails on qemuriscv64 where the process cmdline may not be fully visible in /proc/<pid>/cmdline at the time pgrep reads it, or the large regex match times out under TCG emulation.
Skip this single test case while keeping all other 30+ pgrep tests which are reliable.
[YOCTO #16290]
--- ...p-pgrep-full-process-name-match-test.patch | 39 +++++++++++++++++++ meta/recipes-extended/procps/procps_4.0.6.bb | 1 + 2 files changed, 40 insertions(+) create mode 100644 meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch new file +39 -0
@@ -0,0 +1,39 @@ +From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001 +From: Pratik Farkase <[email protected]> +Date: Mon, 17 Aug 2026 10:00:00 +0000 +Subject: [PATCH] testsuite: skip pgrep full process name match test + +The "pgrep match against full process name" test uses pgrep -f with a +regex pattern close to ARG_MAX in size (~100KB). This intermittently +fails on qemuriscv64 where the process cmdline may not be fully visible +in /proc/<pid>/cmdline at the time pgrep reads it, or the large regex +match times out under TCG emulation. + +Skip this single test case while keeping all other pgrep tests which +are reliable. + +https://bugzilla.yoctoproject.org/show_bug.cgi?id=16290 + +Upstream-Status: Inappropriate [OE-ptest specific: fails only under QEMU TCG emulation] +Signed-off-by: Pratik Farkase <[email protected]> +--- + testsuite/pgrep.test/pgrep.exp | 6 +++--- + 1 file changed, 3 insertions(+), 3 deletions(-) + +diff --git a/testsuite/pgrep.test/pgrep.exp b/testsuite/pgrep.test/pgrep.exp +index 54e75df..66edbaa 100644 +--- a/testsuite/pgrep.test/pgrep.exp ++++ b/testsuite/pgrep.test/pgrep.exp +@@ -35,9 +35,9 @@ set test "pgrep with : delimiter" + spawn $pgrep -d : $testproc_comm + expect_pass "$test" "^${testproc1_pid}:${testproc2_pid}\\s*$" + +-set test "pgrep match against full process name" +-spawn $pgrep -f "$testproc_path\\s+$testproc_arg_str" +-expect_pass "$test" "^$testproc1_pid\\s*$" ++# Skip: intermittent failure on riscv64 - pgrep -f with large ARG_MAX ++# patterns can fail under QEMU emulation (timing/cmdline visibility race) ++untested "pgrep match against full process name" + + set test "pgrep with matching gid" + spawn $pgrep -G $gid $testproc_comm
meta/recipes-extended/procps/procps_4.0.6.bb +1 -0
@@ -16,6 +16,7 @@ SRC_URI = "git://gitlab.com/procps-ng/procps.git;protocol=https;branch=master;ta file://sysctl.conf \ file://0001-tests-Disable-twice-total-pmap-X-tests.patch \ file://0001-testsuite-skip-signal-based-kill-tests.patch \ + file://0001-testsuite-skip-pgrep-full-process-name-match-test.patch \ file://run-ptest \ " SRCREV = "4dafddf4c3f4646caa517f039a2307e92657ec93"
Signature · 1 line
2.43.0
Quoted 4 lines · Pratik Farkase via lists.openembedded.org
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch new file +4
-0+Subject: [PATCH] testsuite: skip pgrep full process name match test + +The "pgrep match against full process name" test uses pgrep -f with a +regex pattern close to ARG_MAX in size (~100KB). This intermittently
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch line 17
Quoted 10 lines
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch new file +10
-0+fails on qemuriscv64 where the process cmdline may not be fully visible +in /proc/<pid>/cmdline at the time pgrep reads it, or the large regex +match times out under TCG emulation. + +Skip this single test case while keeping all other pgrep tests which +are reliable. + +https://bugzilla.yoctoproject.org/show_bug.cgi?id=16290 + +Upstream-Status: Inappropriate [OE-ptest specific: fails only under QEMU TCG emulation]
Hello Pratik,
even though this shows up only under slow software emulation, it's still a race condition in the test, and should be reported upstream, so they can figure out how to sync the test properly. Can you please open a ticket, and update the patch with the link?
Alex
Hello Alex,
Thanks for the review!
It abolutely makes sense i have filed the ticket here : https://gitlab.com/procps-ng/procps/-/work_items/435
Updated the patch with link too here : https://lists.openembedded.org/g/openembedded-core/message/243853
Best Regards,
Pratik
________________________________________
From: Alexander Kanavin <[email protected]>
Sent: Thursday, August 20, 2026 12:28 PM
To: Pratik Farkase
Cc: [email protected]; [email protected]
Subject: Re: [OE-core][PATCH v1] procps: ptest: skip flaky pgrep full process name match test
Quoted 4 lines · Pratik Farkase via lists.openembedded.org
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch new file +4
-0+Subject: [PATCH] testsuite: skip pgrep full process name match test + +The "pgrep match against full process name" test uses pgrep -f with a +regex pattern close to ARG_MAX in size (~100KB). This intermittently
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch line 17
Quoted 10 lines
meta/recipes-extended/procps/procps/0001-testsuite-skip-pgrep-full-process-name-match-test.patch new file +10
-0+fails on qemuriscv64 where the process cmdline may not be fully visible +in /proc/<pid>/cmdline at the time pgrep reads it, or the large regex +match times out under TCG emulation. + +Skip this single test case while keeping all other pgrep tests which +are reliable. + +https://bugzilla.yoctoproject.org/show_bug.cgi?id=16290 + +Upstream-Status: Inappropriate [OE-ptest specific: fails only under QEMU TCG emulation]
Hello Pratik,
even though this shows up only under slow software emulation, it's still a race condition in the test, and should be reported upstream, so they can figure out how to sync the test properly. Can you please open a ticket, and update the patch with the link?
Alex