binutils: fix PV to respect upstream tag and also cve database
cover binutils: fix PV to respect upstream tag and also cve database
*** BLURB HERE ***
The following changes since commit 35742c10f531ba451c06a4e360fecfb55f358b35:
ref-manual: Removed DISK_DESCRIPTION variable from glossary (2019-03-26 15:38:56 +0000)
are available in the git repository at:
git://git.pokylinux.org/poky-contrib ChenQi/binutils-pv
http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=ChenQi/binutils-pv
Chen Qi (1):
binutils: fix PV to respect upstream tag and also cve database
meta/recipes-devtools/binutils/binutils-2.32.inc | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-)
Signature · 1 line
1.9.1
PATCH 1/1 binutils: fix PV to respect upstream tag and also cve database
Setting PV to "2.32.0" results in cve checking tools failing to report the actual unfixed CVEs. We need to change it to "2.32". After the change, when `bitbake -c cve_check binutils-native', we have:
binutils-native-2.32-r0 do_cve_check: Found unpatched CVE (CVE-2018-1000876 CVE-2019-9070 CVE-2019-9071 CVE-2019-9072 CVE-2019-9073 CVE-2019-9074 CVE-2019-9075 CVE-2019-9076 CVE-2019-9077)
Checking the upstream repo, it has tags like:
binutils-2_29
binutils-2_29_1
binutils-2_29_1.1
binutils-2_30
binutils-2_31
binutils-2_31_1
binutils-2_32
So we'd better respect both upstream tags and cve database and use '2.32'. If binutils 2.32.1 is released some day, we then change the PV to '2.32.1'.
--- meta/recipes-devtools/binutils/binutils-2.32.inc | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-)
meta/recipes-devtools/binutils/binutils-2.32.inc +1 -7
@@ -10,13 +10,7 @@ LIC_FILES_CHKSUM="\ file://bfd/COPYING;md5=d32239bcb673463ab874e80d47fae504\ " -def binutils_branch_version(d): - pvsplit = d.getVar('PV').split('.') - return pvsplit[0] + "_" + pvsplit[1] - -PV = "2.32.0" -BINUPV = "${@binutils_branch_version(d)}" -#BRANCH = "binutils-${BINUPV}-branch" +PV = "2.32" BRANCH ?= "binutils-2_32-branch" UPSTREAM_CHECK_GITTAGREGEX = "binutils-(?P<pver>\d+_(\d_?)*)"
Signature · 1 line
1.9.1
Quoted 3 lines · Chen Qi
Setting PV to "2.32.0" results in cve checking tools failing to report the actual unfixed CVEs. We need to change it to "2.32". After the change, when `bitbake -c
cve_checkbinutils-native', we have:
I sent a related patch yesterday:
http://lists.openembedded.org/pipermail/openembedded-core/2019-March/280511.html
The problem with setting PV properly, is that 2.32 is less than 2.32.0 and will cause the 'version going backwards' problem. But if RP/Khem are fine with it, I would prefer your patch.
Alex
Quoted 3 lines · Alexander Kanavin
The problem with setting PV properly, is that 2.32 is less than 2.32.0 and will cause the 'version going backwards' problem. But if RP/Khem are fine with it, I would prefer your patch.
I just had a look at the code in bitbake and yes, it actually even tests that 1 < 1.0. I'd say that there's an argument for handling .0 releases specially and considering 2.32 == 2.32.0.
Ross
Quoted 8 lines · Ross
Quoted 3 lines · Alexander Kanavin
The problem with setting PV properly, is that 2.32 is less than 2.32.0 and will cause the 'version going backwards' problem. But if RP/Khem are fine with it, I would prefer your patch.
I just had a look at the code in bitbake and yes, it actually even tests that 1 < 1.0. I'd say that there's an argument for handling .0 releases specially and considering 2.32 == 2.32.0.
Of course that test exists for people using feeds, and at least dpkg thinks 2.32 < 2.32.0, so even if we changed the test the feed would still be broken.
Ross
Quoted 6 lines · Ross
Quoted 5 lines · Ross
Quoted 3 lines · Alexander Kanavin
The problem with setting PV properly, is that 2.32 is less than 2.32.0 and will cause the 'version going backwards' problem. But if RP/Khem are fine with it, I would prefer your patch.
Quoted 7 lines
Quoted 3 lines
I just had a look at the code in bitbake and yes, it actually even tests that 1 < 1.0. I'd say that there's an argument for handling .0 releases specially and considering 2.32 == 2.32.0.
Of course that test exists for people using feeds, and at least dpkg thinks 2.32 < 2.32.0, so even if we changed the test the feed would still be broken.
Breaking feeds on purpose is bad, so this is a case to pump PE?
Signature · 1 line
Tom
Quoted 9 lines · Tom Rini
Quoted 7 lines
Quoted 3 lines
I just had a look at the code in bitbake and yes, it actually even tests that 1 < 1.0. I'd say that there's an argument for handling .0 releases specially and considering 2.32 == 2.32.0.
Of course that test exists for people using feeds, and at least dpkg thinks 2.32 < 2.32.0, so even if we changed the test the feed would still be broken.
Breaking feeds on purpose is bad, so this is a case to pump PE?
Bumping PE is even worse as a PE never goes away. I say we leave it as it is, and ask upstream nicely if they'll match the tag/version in the future.
The CVE problem can be resolved by setting CVE_VERSION.
Ross
Quoted 3 lines · Ross
Bumping PE is even worse as a PE never goes away. I say we leave it as it is, and ask upstream nicely if they'll match the tag/version in the future.
The tag and version upstream are both 2.32. It's Khem who declared PV to be 2.32.0 :)
Alex
Quoted 8 lines · Alexander Kanavin
Quoted 3 lines · Ross
Bumping PE is even worse as a PE never goes away. I say we leave it as it is, and ask upstream nicely if they'll match the tag/version in the future.
The tag and version upstream are both 2.32. It's Khem who declared PV to be 2.32.0 :)
Bumping PE is what this is for too, to not break package feeds after we've done something wrong in PV. So lets bump PE and not break feeds. Especially since this is so that CVE check starts to see and correctly complain about issues that will result in "lets move this from master to ..." and so break feeds.
Signature · 1 line
Tom
Quoted 5 lines · Tom Rini
Bumping PE is what this is for too, to not break package feeds after we've done something wrong in PV. So lets bump PE and not break feeds. Especially since this is so that CVE check starts to see and correctly complain about issues that will result in "lets move this from master to ..." and so break feeds.
Bumping PE is a nuclear-armed hammer when setting CVE_VERSION="2.32" is sufficient and will go away when we upgrade to 2.32.1 onwards.
Ross
Quoted 9 lines · Ross
Quoted 5 lines · Tom Rini
Bumping PE is what this is for too, to not break package feeds after we've done something wrong in PV. So lets bump PE and not break feeds. Especially since this is so that CVE check starts to see and correctly complain about issues that will result in "lets move this from master to ..." and so break feeds.
Bumping PE is a nuclear-armed hammer when setting CVE_VERSION="2.32" is sufficient and will go away when we upgrade to 2.32.1 onwards.
So long as we don't break package feeds, OK. Thanks!
Signature · 1 line
Tom
Quoted 2 lines · Alexander Kanavin
The tag and version upstream are both 2.32. It's Khem who declared PV to be 2.32.0 :)
Yes I noticed that after sending. Let's ask Khem nicely to use the right version. :)
Ross