Bug#1021310: libsdl2: FTBFS on hppa - testevdev: FAILED: 1

2022-10-05 Thread John David Anglin

On 2022-10-05 11:11 a.m., Simon McVittie wrote:

It seems to be reliable on release architectures, including i386 which is
32-bit and s390x which is big-endian.

Seems to be a 32-bit big-endian issue as same fail occurs on powerpc but not 
ppc64 or ppc64el.
Possibly, wrong word in 64-bit type is being tested on hppa and powerpc.

Dave

--
John David Anglin  dave.ang...@bell.net



Bug#1021310: libsdl2: FTBFS on hppa - testevdev: FAILED: 1

2022-10-05 Thread Simon McVittie
Control: severity -1 normal
Control: tags -1 + help

On Wed, 05 Oct 2022 at 14:43:00 +, John David Anglin wrote:
> Justification: fails to build from source (but built successfully in the past)

hppa is not a release architecture, so FTBFS on hppa is not RC.

This test-case seems to have been failing on hppa around 75% of the time
since it was added. I don't know why it would do that, because this
test-case should be deterministic: it's inspecting a simulation of the
capabilitities info that we would have received from the kernel for a
particular input device, and running SDL's algorithm to guess what type
of input device it represents (keyboard, mouse, gamepad, etc.). When it
fails on hppa, the failure mode is that none of the bits representing a
device type get set.

It seems to be reliable on release architectures, including i386 which is
32-bit and s390x which is big-endian.

Non-release architecture porters are welcome to investigate why this
is going wrong on their favourite architecture, and if the root cause
turns out to be a bug in SDL, I'm sure upstream would appreciate a patch.

smcv



Bug#1021310: libsdl2: FTBFS on hppa - testevdev: FAILED: 1

2022-10-05 Thread John David Anglin
Source: libsdl2
Version: 2.24.0+dfsg-2
Severity: serious
Tags: ftbfs
Justification: fails to build from source (but built successfully in the past)

Dear Maintainer,

Build fails in testsuite:

Thinkpad USB keyboard with Trackpoint - Trackpoint...
Expected 0x0003
MOUSE
KEYBOARD
Got  0x
No information...
OK
testevdev: FAILED: 1
testfilesystem...
INFO: base path: '/<>/debian/build-tests/'
INFO: pref path: 
'/<>/debian/.debhelper/generated/_source/home/.local/share/libsdl/test_filesystem/'
INFO: pref path: 
'/<>/debian/.debhelper/generated/_source/home/.local/share/test_filesystem/'
testfilesystem: OK
...
testdisplayinfo: OK
make[2]: *** [Makefile:416: check] Error 1
make[2]: Leaving directory '/<>/debian/build-tests'
dh_auto_test: error: cd debian/build-tests && make -j4 check 
"TESTSUITEFLAGS=-j4 --verbose" VERBOSE=1 V=1 returned exit code 2
make[1]: *** [debian/rules:142: override_dh_auto_test-arch] Error 25
make[1]: Leaving directory '/<>'
make: *** [debian/rules:81: binary-arch] Error 2
dpkg-buildpackage: error: debian/rules binary-arch subprocess returned exit 
status 2

Full log is here:
https://buildd.debian.org/status/fetch.php?pkg=libsdl2=hppa=2.24.1%2Bdfsg-1=1664977752=0

Similar fail occurs on powerpc.

Regards,
Dave Anglin

-- System Information:
Debian Release: bookworm/sid
  APT prefers buildd-unstable
  APT policy: (500, 'buildd-unstable'), (500, 'unstable')
Architecture: hppa (parisc64)

Kernel: Linux 5.19.13+ (SMP w/4 CPU threads)
Kernel taint flags: TAINT_WARN
Locale: LANG=C, LC_CTYPE=C.UTF-8 (charmap=UTF-8), LANGUAGE not set
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)