Skip to contentSkip to conversation

bitbake.conf: Add timeout to HOSTTOOLS

PATCH In review [email protected] 3 participants 3 messages Aug 7 – Aug 10

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 bitbake.conf: Add timeout to HOSTTOOLS

From: Hemanth Kumar M D <[email protected]>

The glibc iconv/tst-iconv_prog test uses timeout(1) on the host to guard against hangs during charset conversion testing. Since timeout is not listed in HOSTTOOLS, it is unavailable in the Yocto build environment, causing the test to fail with exit code 127 (command not found) instead of the expected conversion error code.

AB testresults (qemux86-64-tc):

  +--------------+--------+--------+------+
  | Result       | Before | After  | Diff |
  +--------------+--------+--------+------+
  | PASS         | 6759   | 6760   | +1   |
  | FAIL         | 30     | 29     | -1   |
  | UNSUPPORTED  | 584    | 584    |  0   |
  | XFAIL        | 16     | 16     |  0   |
  +--------------+--------+--------+------+

iconv/tst-iconv_prog FAIL -> PASS

With this fix, iconv/tst-iconv_prog now passes on all architectures. Verified on AB.

---
 meta/conf/bitbake.conf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
meta/conf/bitbake.conf +1 -1
index bdf37d0da2..39966d60ef 100644
@@ -533,7 +533,7 @@ HOSTTOOLS += " \
     mktemp mv nm objcopy objdump od patch perl pr printf pwd \
     python3 pzstd ranlib readelf readlink realpath rm rmdir rpcgen sed seq sh \
     sha1sum sha224sum sha256sum sha384sum sha512sum \
-    sleep sort split stat strings strip tail tar tee test touch tr true truncate uname \
+    sleep sort split stat strings strip tail tar tee test timeout touch tr true truncate uname \
     uniq unzstd wc wget which xargs zstd xz \
 "
 
Signature · 1 line
2.49.0
#1 lore raw
Quoted 8 lines · Hemanth Kumar M D via lists.openembedded.org

The glibc iconv/tst-iconv_prog test uses timeout(1) on the host to guard against hangs during charset conversion testing. Since timeout is not listed in HOSTTOOLS, it is unavailable in the Yocto build environment, causing the test to fail with exit code 127 (command not found) instead of the expected conversion error code.

meta/conf/bitbake.conf HOSTTOOLS += " \ · line 538

Quoted 7 lines
meta/conf/bitbake.conf +1 -1
@@ -533,7 +533,7 @@ HOSTTOOLS += " \
     mktemp mv nm objcopy objdump od patch perl pr printf pwd \
     python3 pzstd ranlib readelf readlink realpath rm rmdir rpcgen sed seq sh \
     sha1sum sha224sum sha256sum sha384sum sha512sum \
-    sleep sort split stat strings strip tail tar tee test touch tr true truncate uname \
+    sleep sort split stat strings strip tail tar tee test timeout touch tr true truncate uname \
     uniq unzstd wc wget which xargs zstd xz \
 "

Adding to hosttools just to fix a single test is very excessive. Rather make it depend on coreutils-native.

Alex

#2 lore raw
Quoted 13 lines · Alexander Kanavin via lists.openembedded.org
Quoted 10 lines · Hemanth Kumar M D via lists.openembedded.org

The glibc iconv/tst-iconv_prog test uses timeout(1) on the host to guard against hangs during charset conversion testing. Since timeout is not listed in HOSTTOOLS, it is unavailable in the Yocto build environment, causing the test to fail with exit code 127 (command not found) instead of the expected conversion error code.

meta/conf/bitbake.conf +0 -0
@@ -533,7 +533,7 @@ HOSTTOOLS += " \
      mktemp mv nm objcopy objdump od patch perl pr printf pwd \
      python3 pzstd ranlib readelf readlink realpath rm rmdir rpcgen sed seq sh \
Quoted 8 lines
Quoted 5 lines
meta/conf/bitbake.conf +1 -1
      sha1sum sha224sum sha256sum sha384sum sha512sum \
-    sleep sort split stat strings strip tail tar tee test touch tr true truncate uname \
+    sleep sort split stat strings strip tail tar tee test timeout touch tr true truncate uname \
      uniq unzstd wc wget which xargs zstd xz \

 "

Adding to hosttools just to fix a single test is very excessive. Rather make it depend on coreutils-native.

I'm a bit torn on this. Since timeout is part of coreutils, the host dependencies are basically unchanged by the patch since we already require a good chunk of coreutils.

I do wish we could make "per recipe" markup of things like this work in a reasonable way though as it is sad to mark this as needed everywhere...

Cheers,

Richard