Richard Purdie : sstate.bbclass: Detect overwriting files in sstate controlled directories
Quoted 117 lines · On Fri, Aug 31, 2012 at 2:38 AM, <[email protected]> wrote:
Module: openembedded-core.git Branch: master Commit: 1f1b6f93d6b7aa8c9bd9bb5b1826997812e36932 URL: http://git.openembedded.org/?p=openembedded-core.git&a=commit;h=1f1b6f93d6b7aa8c9bd9bb5b1826997812e36932Author: Richard Purdie <[email protected]> Date: Sat Aug 25 15:48:14 2012 +0000sstate.bbclass: Detect overwriting files in sstate controlled directories
We have a potential problem where two sstate packages try and touch the same file. This adds code which will print a warning whenever this happens.
The implementation does but by maintaining a master file list and comparing file accesses against this. There are a number of places we have duplicate accesses which are harmless, mostly in the deploy directory so these are whitelisted.
For now the code prints warnings, this could be strengthened in future to become error messages. Whilst working on this code on and off over the past few months various issues were uncovered, some serious.
[YOCTO #238]
Signed-off-by Richard Purdie <[email protected]>---
meta/classes/sstate.bbclass | 45 ++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 44 insertions(+), 1 deletions(-)meta/classes/sstate.bbclass +37
-1@@ -3,6 +3,7 @@ SSTATE_VERSION = "2" SSTATE_MANIFESTS ?= "${TMPDIR}/sstate-control" SSTATE_MANFILEBASE = "${SSTATE_MANIFESTS}/manifest-${SSTATE_MANMACH}-" SSTATE_MANFILEPREFIX = "${SSTATE_MANFILEBASE}${PN}" +SSTATE_MASTERMANIFEST = "${SSTATE_MANIFESTS}/master.list" def generate_sstatefn(spec, hash, d): if not hash:@@ -17,6 +18,7 @@ SSTATE_EXTRAPATH = "" SSTATE_EXTRAPATHWILDCARD = "" SSTATE_PATHSPEC = "${SSTATE_DIR}/${SSTATE_EXTRAPATHWILDCARD}*/${SSTATE_PKGSPEC}" +SSTATE_DUPWHITELIST = "${DEPLOY_DIR_IMAGE}/ ${DEPLOY_DIR}/licenses/ ${DEPLOY_DIR_IPK}/all/ ${DEPLOY_DIR_RPM}/all ${DEPLOY_DIR_DEB}/all/" SSTATE_SCAN_FILES ?= "*.la *-config *_config" SSTATE_SCAN_CMD ?= 'find ${SSTATE_BUILDDIR} \( -name "${@"\" -o -name \"".join(d.getVar("SSTATE_SCAN_FILES", True).split())}" \) -type f'@@ -125,7 +127,6 @@ def sstate_install(ss, d): locks.append(bb.utils.lockfile(lock)) for state in ss['dirs']: - oe.path.copytree(state[1], state[2]) for walkroot, dirs, files in os.walk(state[1]): bb.debug(2, "Staging files from %s to %s" % (state[1], state[2])) for file in files:@@ -140,9 +141,35 @@ def sstate_install(ss, d): if not dstdir.endswith("/"): dstdir = dstdir + "/" shareddirs.append(dstdir) + + # Check the file list for conflicts against the master manifest + mastermanifest = d.getVar("SSTATE_MASTERMANIFEST", True) + whitelist = d.getVar("SSTATE_DUPWHITELIST", True) + lock = bb.utils.lockfile(mastermanifest + ".lock") + fileslist = [line.strip() for line in open(mastermanifest)] + bb.utils.unlockfile(lock) + match = [] + for f in sharedfiles: + if f in fileslist: + realmatch = True + for w in whitelist: + if f.startswith(w): + realmatch = False + break + if realmatch: + match.append(f) + if match: + bb.warn("The recipe is trying to install files into a shared area when those files already exist. Those files are:\n %s" % "\n ".join(match)) + + # Write out the manifest and add to the task's manifest file + lock = bb.utils.lockfile(mastermanifest + ".lock") + mf = open(mastermanifest, "a") f = open(manifest, "w") for file in sharedfiles: + mf.write(file + "\n") f.write(file + "\n") + bb.utils.unlockfile(lock) + # We want to ensure that directories appear at the end of the manifest # so that when we test to see if they should be deleted any contents # added by the task will have been removed first.@@ -152,6 +179,10 @@ def sstate_install(ss, d): f.write(di + "\n") f.close() + # Run the actual file install + for state in ss['dirs']: + oe.path.copytree(state[1], state[2]) + for postinst in (d.getVar('SSTATEPOSTINSTFUNCS', True) or '').split(): bb.build.exec_func(postinst, d)@@ -268,6 +299,18 @@ def sstate_clean_manifest(manifest, d): except OSError: pass + # Remove the entries from the master manifest + mastermanifest = d.getVar("SSTATE_MASTERMANIFEST", True) + lock = bb.utils.lockfile(mastermanifest + ".lock") + mf = open(mastermanifest + ".new", "w") + for line in open(mastermanifest, "r"):
meta/classes/sstate.bbclass def sstate_clean_manifest(manifest, d): · line 316
Quoted 10 lines
meta/classes/sstate.bbclass +7
-0+ if not line or line in entries: + continue + mf.write(line) + mf.close() + os.rename(mastermanifest + ".new", mastermanifest) + bb.utils.unlockfile(lock) + oe.path.remove(manifest) def sstate_clean(ss, d):
I think it started with this one: I often do play around with different kernel versions/sources to bisect when a bug was born. Doing so I now get wallpapers with
WARNING: The recipe is trying to install files into a shared area when those files already exist. Those files are:
/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/Module.symvers /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/CREDITS /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/kernel-abiversion /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/Makefile /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/kernel-image-name /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/uImage /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/README /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/COPYING /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/REPORTING-BUGS /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/System.map-3.5.0
...
(log.do_populate_sysroot is 2.8 MB)
Would it cause huge effort to add a variable which is in scope of recipe avoiding this warning because it scrolls out all other interesting messages?
Andreas
Quoted 116 lines · On Fri, Aug 31, 2012 at 2:38 AM, <[email protected]> wrote:
Module: openembedded-core.git Branch: master Commit: 1f1b6f93d6b7aa8c9bd9bb5b1826997812e36932 URL: http://git.openembedded.org/?p=openembedded-core.git&a=commit;h=1f1b6f93d6b7aa8c9bd9bb5b1826997812e36932Author: Richard Purdie <[email protected]> Date: Sat Aug 25 15:48:14 2012 +0000sstate.bbclass: Detect overwriting files in sstate controlled directories
We have a potential problem where two sstate packages try and touch the same file. This adds code which will print a warning whenever this happens.
The implementation does but by maintaining a master file list and comparing file accesses against this. There are a number of places we have duplicate accesses which are harmless, mostly in the deploy directory so these are whitelisted.
For now the code prints warnings, this could be strengthened in future to become error messages. Whilst working on this code on and off over the past few months various issues were uncovered, some serious.
[YOCTO #238]
Signed-off-by Richard Purdie <[email protected]>---
meta/classes/sstate.bbclass | 45 ++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 44 insertions(+), 1 deletions(-)meta/classes/sstate.bbclass +37
-1@@ -3,6 +3,7 @@ SSTATE_VERSION = "2" SSTATE_MANIFESTS ?= "${TMPDIR}/sstate-control" SSTATE_MANFILEBASE = "${SSTATE_MANIFESTS}/manifest-${SSTATE_MANMACH}-" SSTATE_MANFILEPREFIX = "${SSTATE_MANFILEBASE}${PN}" +SSTATE_MASTERMANIFEST = "${SSTATE_MANIFESTS}/master.list" def generate_sstatefn(spec, hash, d): if not hash:@@ -17,6 +18,7 @@ SSTATE_EXTRAPATH = "" SSTATE_EXTRAPATHWILDCARD = "" SSTATE_PATHSPEC = "${SSTATE_DIR}/${SSTATE_EXTRAPATHWILDCARD}*/${SSTATE_PKGSPEC}" +SSTATE_DUPWHITELIST = "${DEPLOY_DIR_IMAGE}/ ${DEPLOY_DIR}/licenses/ ${DEPLOY_DIR_IPK}/all/ ${DEPLOY_DIR_RPM}/all ${DEPLOY_DIR_DEB}/all/" SSTATE_SCAN_FILES ?= "*.la *-config *_config" SSTATE_SCAN_CMD ?= 'find ${SSTATE_BUILDDIR} \( -name "${@"\" -o -name \"".join(d.getVar("SSTATE_SCAN_FILES", True).split())}" \) -type f'@@ -125,7 +127,6 @@ def sstate_install(ss, d): locks.append(bb.utils.lockfile(lock)) for state in ss['dirs']: - oe.path.copytree(state[1], state[2]) for walkroot, dirs, files in os.walk(state[1]): bb.debug(2, "Staging files from %s to %s" % (state[1], state[2])) for file in files:@@ -140,9 +141,35 @@ def sstate_install(ss, d): if not dstdir.endswith("/"): dstdir = dstdir + "/" shareddirs.append(dstdir) + + # Check the file list for conflicts against the master manifest + mastermanifest = d.getVar("SSTATE_MASTERMANIFEST", True) + whitelist = d.getVar("SSTATE_DUPWHITELIST", True) + lock = bb.utils.lockfile(mastermanifest + ".lock") + fileslist = [line.strip() for line in open(mastermanifest)] + bb.utils.unlockfile(lock) + match = [] + for f in sharedfiles: + if f in fileslist: + realmatch = True + for w in whitelist: + if f.startswith(w): + realmatch = False + break + if realmatch: + match.append(f) + if match: + bb.warn("The recipe is trying to install files into a shared area when those files already exist. Those files are:\n %s" % "\n ".join(match)) + + # Write out the manifest and add to the task's manifest file + lock = bb.utils.lockfile(mastermanifest + ".lock") + mf = open(mastermanifest, "a") f = open(manifest, "w") for file in sharedfiles: + mf.write(file + "\n") f.write(file + "\n") + bb.utils.unlockfile(lock) + # We want to ensure that directories appear at the end of the manifest # so that when we test to see if they should be deleted any contents # added by the task will have been removed first.@@ -152,6 +179,10 @@ def sstate_install(ss, d): f.write(di + "\n") f.close() + # Run the actual file install + for state in ss['dirs']: + oe.path.copytree(state[1], state[2]) + for postinst in (d.getVar('SSTATEPOSTINSTFUNCS', True) or '').split(): bb.build.exec_func(postinst, d)@@ -268,6 +299,18 @@ def sstate_clean_manifest(manifest, d): except OSError: pass + # Remove the entries from the master manifest + mastermanifest = d.getVar("SSTATE_MASTERMANIFEST", True) + lock = bb.utils.lockfile(mastermanifest + ".lock") + mf = open(mastermanifest + ".new", "w") + for line in open(mastermanifest, "r"):
meta/classes/sstate.bbclass def sstate_clean_manifest(manifest, d): · line 316
Quoted 10 lines
meta/classes/sstate.bbclass +7
-0+ if not line or line in entries: + continue + mf.write(line) + mf.close() + os.rename(mastermanifest + ".new", mastermanifest) + bb.utils.unlockfile(lock) + oe.path.remove(manifest) def sstate_clean(ss, d):
This previous addressed the wrong list - so again:
I think it started with this one: I often do play around with different kernel versions/sources to bisect when a bug was born. Doing so I now get wallpapers with
WARNING: The recipe is trying to install files into a shared area when those files already exist. Those files are:
/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/Module.symvers /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/CREDITS /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/kernel-abiversion /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/Makefile /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/kernel-image-name /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/uImage /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/README /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/COPYING /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/REPORTING-BUGS /home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/System.map-3.5.0
...
(log.do_populate_sysroot is 2.8 MB)
Would it cause huge effort to add a variable which is in scope of recipe avoiding this warning because it scrolls out all other interesting messages?
Andreas
Quoted 21 lines · Andreas Müller
This previous addressed the wrong list - so again:
I think it started with this one: I often do play around with different kernel versions/sources to bisect when a bug was born. Doing so I now get wallpapers with
WARNING: The recipe is trying to install files into a shared area when those files already exist. Those files are:
/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/Module.symvers/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/CREDITS/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/kernel-abiversion/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/Makefile/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/kernel-image-name/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/uImage/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/README/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/COPYING/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/REPORTING-BUGS/home/Superandy/tmp/oe-core-eglibc/sysroots/overo/usr/src/kernel/System.map-3.5.0
...
(log.do_populate_sysroot is 2.8 MB)
Quoted 3 lines
Would it cause huge effort to add a variable which is in scope of recipe avoiding this warning because it scrolls out all other interesting messages?
There was some misconfiguration of the whitelist, this should be a lot less noisy now.
Cheers,
Richard