Fix grep pattern when mklibs collects executables in rootfs
PATCH 2/2 Fix grep pattern when mklibs collects executables in rootfs
From: Lei Liu <[email protected]>
File command in some version could print extra space between "LSB" and "executable" - it causes mklibs can't find any executables using grep "LSB executable". Fix the grep pattern to catch multiple spaces.
--- meta/classes/image-mklibs.bbclass | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
meta/classes/image-mklibs.bbclass +1 -1
@@ -9,7 +9,7 @@ mklibs_optimize_image_doit() { du -bs > ${WORKDIR}/mklibs/du.before.mklibs.txt for i in `find .`; do file $i; done \ | grep ELF \ - | grep "LSB executable" \ + | grep "LSB *executable" \ | grep "dynamically linked" \ | sed "s/:.*//" \ | sed "s+^\./++" \
Signature · 1 line
1.7.0.4
Quoted 11 lines · Lei Liu
From: Lei Liu <[email protected]>
File command in some version could print extra space between "LSB" and "executable" - it causes mklibs can't find any executables using grep "LSB executable". Fix the grep pattern to catch multiple spaces.
Signed-off-by Lei Liu <[email protected]>--- meta/classes/image-mklibs.bbclass | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
I've queued this but please put the area name in the short line of the commit in future ("image-mklibs: " in this case).
Cheers,
Richard
Quoted 8 lines · Richard Purdie
Quoted 7 lines · Lei Liu
From: Lei Liu <[email protected]>
File command in some version could print extra space between "LSB" and "executable" - it causes mklibs can't find any executables using grep "LSB executable". Fix the grep pattern to catch multiple spaces.
Quoted 7 lines
Quoted 4 lines
Signed-off-by Lei Liu <[email protected]>--- meta/classes/image-mklibs.bbclass | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)I've queued this but please put the area name in the short line of the commit in future ("image-mklibs: " in this case).
Okay. Thanks.
Lei
hi,
Quoted 13 lines · Lei Liu
Quoted 12 lines · Richard Purdie
Quoted 11 lines · Lei Liu
From: Lei Liu <[email protected]>
File command in some version could print extra space between "LSB" and "executable" - it causes mklibs can't find any executables using grep "LSB executable". Fix the grep pattern to catch multiple spaces.
Signed-off-by Lei Liu <[email protected]>--- meta/classes/image-mklibs.bbclass | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
Quoted 5 lines
Quoted 3 lines
I've queued this but please put the area name in the short line of the commit in future ("image-mklibs: " in this case).
Okay. Thanks.
First of all, this patch is needed on dora branch too , and it applies cleanly, can that be merged?
Now, the real question... i am not able to get mklibs to work at all (even with this patch). What i am seeing is that mklibs 0.1.38 doesn't support anymore the use of 'sysroot'. I think this is because of this upstream change
If I revert to 0.1.34 (that we had on dylan branch), i can build my image just fine (note that I am building an arm soft-float image), so are you able to use image-mklibs with dora or master which have 0.1.38?
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 10 lines · Lei Liu
Quoted 1 line
If I revert to 0.1.34 (that we had on dylan branch), i can build my
image just fine (note that I am building an arm soft-float image), so are you able to use image-mklibs with dora or master which have 0.1.38?
Quoted 1 line
Hi,
I'm testing on master. And I'm able to build my qemux86 image with mklibs-0.1.38.
ok, thanks for your answer.
I tested this morning, and confirm that 'master' + mklibs is broken on ARM (i tested qemuarm), altough qemux86 is working.. i will try to look into that... in fact to be honest, i still believe that mklibs 0.1.38 is broken, even on x86 and that it's working for qemux86 because it uses the loader from your host instead of the one in the sysroot. In my case it's failing because it doesn't find the loader in /lib (/lib in my host):
Exception: Cannot find lib /lib/ld-linux.so.3
it should in theory look for that file in the sysroot... if you look at the upstream commit i referenced above, upstream no longer uses 'sysroot' variable to locate the loader...