Skip to contentSkip to conversation

selftests/bpf: workaround stdout issue in VM launched by vmtest.sh

PATCH In review Yucong Sun 3 participants 3 messages Feb 23, 2022

Subsystems (3)
lore
Share & tools
lore

Yucong Sun

author
#0 lore raw

PATCH 1/1 selftests/bpf: workaround stdout issue in VM launched by vmtest.sh

This apply a workaround to fix stdout issue in ./vmtest.sh invocations, but doesn't work on ./vmtest.sh -s

---
 tools/testing/selftests/bpf/vmtest.sh | 4 ++++
 1 file changed, 4 insertions(+)
tools/testing/selftests/bpf/vmtest.sh +4 -0
index e0bb04a97e10..a9f943a84ed5 100755
@@ -184,6 +184,10 @@ EOF
 	fi
 
 	sudo bash -c "echo '#!/bin/bash' > ${init_script}"
+	sudo bash -c "cat >>${init_script}" <<EOF
+# Force rebinding stdout/stderr to /dev/ttyS0, to workaround a mysterious issue
+exec 1>/dev/ttyS0 2>/dev/ttyS0
+EOF
 
 	if [[ "${command}" != "" ]]; then
 		sudo bash -c "cat >>${init_script}" <<EOF
Signature · 1 line
2.30.2
reviewer
#1 lore raw
Quoted 5 lines · Yucong Sun

This apply a workaround to fix stdout issue in ./vmtest.sh invocations, but doesn't work on ./vmtest.sh -s

Could you please provide more information about the issue?

Quoted 22 lines
---
 tools/testing/selftests/bpf/vmtest.sh | 4 ++++
 1 file changed, 4 insertions(+)
tools/testing/selftests/bpf/vmtest.sh +4 -1
index e0bb04a97e10..a9f943a84ed5 100755
@@ -184,6 +184,10 @@ EOF
        fi
 
        sudo bash -c "echo '#!/bin/bash' > ${init_script}"
+       sudo bash -c "cat >>${init_script}" <<EOF
+# Force rebinding stdout/stderr to /dev/ttyS0, to workaround a mysterious issue
+exec 1>/dev/ttyS0 2>/dev/ttyS0
+EOF
 
        if [[ "${command}" != "" ]]; then
                sudo bash -c "cat >>${init_script}" <<EOF
--

2.30.2

#2 lore raw
Quoted 1 line

Could you please provide more information about the issue?

Still trying to get to the root cause of this issue, I've found that
after https://github.com/kernel-patches/bpf/commit/cb80ddc67152e72f28ff6ea8517acdf875d7381d
, the init process would lose stdout on startup .

ls -lah /proc/1/fd/0 /proc/1/fd/1 /proc/1/fd/2 /proc/1/fd/6 /proc/1/fd/7 lrwx------1 root root 64 Feb 23 17:20 /proc/1/fd/0 -> /dev/null lrwx------ 1 root root 64 Feb 23 17:20 /proc/1/fd/1 -> /dev/null lrwx------ 1 root root 64 Feb 23 17:20 /proc/1/fd/2 -> /dev/console lrwx------ 1 root root 64 Feb 23 17:20 /proc/1/fd/6 -> /dev/console lrwx------ 1 root root 64 Feb 23 17:20 /proc/1/fd/7 -> /dev/console

and same command on the good commit
ls -lah /proc/1/fd/0 /proc/1/fd/1 /proc/1/fd/2 lrwx------ 1 root root 64 Feb 23 17:23 /proc/1/fd/0 -> /dev/console lrwx------ 1 root root 64 Feb 23 17:23 /proc/1/fd/1 -> /dev/console lrwx------ 1 root root 64 Feb 23 17:23 /proc/1/fd/2 -> /dev/console

This patch is merely a workaround until we find and fix the real issue.

Cheers.