Skip to contentSkip to conversation

io_tests fuse fixes

PATCH Superseded Eric Farman 2 participants 6 messages Jul 25 – Jul 28

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover io_tests fuse fixes

Hi,

The following are three small fixes against some of the qemu iotests to avoid some errors that should prevent the tests from running in the first place. We noticed these on an s390 host, but I don't believe that to be relevant. My hunch is the messages are slightly different based on distro, because the tests (correctly) skip for me running Fedora on s390 or x86_64 while they fail under Ubuntu 26.04.

Eric Farman (3):
iotests/108: skip test if cannot access /dev/fuse iotests/file-io-error: complete permission denied check iotests/fuse-mmap-shared: complete permission denied check

 tests/qemu-iotests/108                    | 2 +-
 tests/qemu-iotests/tests/file-io-error    | 3 +++
 tests/qemu-iotests/tests/fuse-mmap-shared | 7 +++++--
 3 files changed, 9 insertions(+), 3 deletions(-)
Signature · 1 line
2.53.0
#1 lore raw

PATCH 1/3 iotests/108: skip test if cannot access /dev/fuse

This test might fail if the user is not root (or sudo'd):

fusermount3: failed to open .../scratch/qcow2-file-108/fuse-export: Permission denied

A check exists to try to cover for this, but there are two potential error messages that signal this case (per the comment block associated with the check) and only a skip for one of them.

Add a check for the other potential error message.

Fixes 9ffd6d646d ("iotests/108: Test new refcount rebuild algorithm")
---
 tests/qemu-iotests/108 | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/qemu-iotests/108 +1 -1
index bf808b3512..136007ceaf 100755
@@ -69,7 +69,7 @@ else
     # or "Invalid parameter 'fuse'", depending on whether there is
     # FUSE support or not.
     error=$($QSD --export fuse 2>&1)
-    if [[ $error = *"'fuse'"* ]]; then
+    if [[ $error = *"'fuse'"* || $error = *"'id'"* ]]; then
         _notrun 'Passwordless sudo for losetup or FUSE support required, but' \
                 'neither is available'
     fi
Signature · 1 line
2.53.0
#2 lore raw

PATCH 2/3 iotests/file-io-error: complete permission denied check

In reply to Eric Farman

As with iotests/108, there are multiple potential error messages returned to the user. Add a 'notrun' exit for another possible error that would otherwise cause this test to fail:

fusermount3: failed to open TEST_DIR/fuse-export: Permission denied

Fixes 380448464d ("tests/file-io-error: New test")
---
 tests/qemu-iotests/tests/file-io-error | 3 +++
 1 file changed, 3 insertions(+)
tests/qemu-iotests/tests/file-io-error +3 -0
index fb8db73b31..86f8e1b6a6 100755
@@ -92,6 +92,9 @@ output=$(_send_qemu_cmd $QEMU_HANDLE \
 if echo "$output" | grep -q "Parameter 'type' does not accept value 'fuse'"; then
     _notrun 'No FUSE support'
 fi
+if echo "$output" | grep -q "failed to open TEST_DIR/fuse-export: Permission denied"; then
+    _notrun 'No FUSE support'
+fi
 echo "$output"
 
 echo
Signature · 1 line
2.53.0
#3 lore raw

PATCH 3/3 iotests/fuse-mmap-shared: complete permission denied check

In reply to Eric Farman

As with iotests/108, there exists multiple messages that would signal a permission error when accessing the fuse tests:

fusermount3: failed to open .../scratch/qcow2-file-fuse-mmap-shared/export.fuse: Permission denied

Incorporate an additional potential error message to cause the test to be skipped rather than failed.

Fixes ba20257d9b ("iotests: test shared mmap for fuse export")
---
 tests/qemu-iotests/tests/fuse-mmap-shared | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
tests/qemu-iotests/tests/fuse-mmap-shared +5 -2
index 52941a3bb6..faa302e4db 100755
@@ -28,8 +28,11 @@ def test_fuse_support(mount_point):
     })
     test_qsd.stop()
     if 'error' in res:
-        assert (res['error']['desc'] ==
-                "Parameter 'type' does not accept value 'fuse'")
+        print(res)
+        assert ((res['error']['desc'] ==
+                "Parameter 'type' does not accept value 'fuse'") or
+                (res['error']['desc'] ==
+                "Failed to mount FUSE session to export"))
         iotests.notrun('No FUSE support')
 
 # Shared mmap when using direct IO is only supported for Linux kernels >= 6.6
Signature · 1 line
2.53.0
#4 lore raw
In reply to Eric Farman

Am 25.07.2026 um 14:56 hat Eric Farman geschrieben:

Quoted 8 lines

Hi,

The following are three small fixes against some of the qemu iotests to avoid some errors that should prevent the tests from running in the first place. We noticed these on an s390 host, but I don't believe that to be relevant. My hunch is the messages are slightly different based on distro, because the tests (correctly) skip for me running Fedora on s390 or x86_64 while they fail under Ubuntu 26.04.

This series conflicts with a patch from Denis Lunev I already queued:

[PATCH v2 2/3] iotests: skip FUSE tests when FUSE is not usable
https://patchew.org/QEMU/[email protected]/[email protected]/

Can you please check if this already solves some of your problems and rebase the rest of your series on top of it?

Kevin

#5 lore raw
Quoted 10 lines · Kevin Wolf

Am 25.07.2026 um 14:56 hat Eric Farman geschrieben:

Quoted 8 lines

Hi,

The following are three small fixes against some of the qemu iotests to avoid some errors that should prevent the tests from running in the first place. We noticed these on an s390 host, but I don't believe that to be relevant. My hunch is the messages are slightly different based on distro, because the tests (correctly) skip for me running Fedora on s390 or x86_64 while they fail under Ubuntu 26.04.

Quoted 7 lines

This series conflicts with a patch from Denis Lunev I already queued:

[PATCH v2 2/3] iotests: skip FUSE tests when FUSE is not usable
https://patchew.org/QEMU/[email protected]/[email protected]/

Can you please check if this already solves some of your problems and rebase the rest of your series on top of it?

Ah, thanks for the pointer. That does indeed cover some of them, but not 108. I'll send that by itself as a v2 shortly.

Quoted 3 lines

Kevin