Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
On Wed, Sep 15, 2021 at 5:43 AM Ranjitsinh Rathod < ranjitsinh.rat...@kpit.com> wrote: > Hi Steve, > > If you wanted to take changes only for the > 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > then you can cherry-pick it from master as I have submitted it for master > and it is available on master branch now. Below is the link. > poky - Poky Build Tool and Metadata (yoctoproject.org) > <https://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=9886ef691aa117d67e4342c6a5e3f79f6a05f8d5> > > Do you still want me to send v2 patch here? > No need, I'll cherry-pick the patch from master. Thanks! Steve > > Thanks, > > Best Regards, > > *Ranjitsinh Rathod* > Technical Leader | | KPIT Technologies Ltd. > Cellphone: +91-84606 92403 > > *__ *KPIT <http://www.kpit.com/> | > Follow us on LinkedIn <http://www.kpit.com/linkedin> > > <https://www.kpit.com/TheNewBrand> > -- > *From:* openembedded-core@lists.openembedded.org < > openembedded-core@lists.openembedded.org> on behalf of Alexander Kanavin > via lists.openembedded.org > *Sent:* Wednesday, September 15, 2021 8:36 PM > *To:* Steve Sakoman > *Cc:* Ranjitsinh Rathod ; Patches and > discussions about the oe-core layer < > openembedded-core@lists.openembedded.org>; Ranjitsinh Rathod < > ranjitsinh.rat...@kpit.com> > *Subject:* Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return > value to avoid major issues and removing unnecessary code > > Caution: This email originated from outside of the KPIT. Do not click > links or open attachments unless you recognize the sender and know the > content is safe. > At this point I have to note that I am removing the patch altogether with > the upcoming upgrade of rpm to 4.17, as I'm also switching the compression > format to zstd, and the patch is generally difficult to maintain and > rebase. If you care about xz compression, please do work with upstream to > get it merged there. > > Alex > > On Wed, 15 Sept 2021 at 16:59, Steve Sakoman wrote: > > On Wed, Sep 8, 2021 at 4:02 AM Ranjitsinh Rathod > wrote: > > > > From: Ranjitsinh Rathod > > > > Change in 2 patch as below to avoid critical issues > > 1) 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > Handled return values of getrlimit() and lzma_cputhreads() functions > > to avoid unexpected behaviours like devide by zero and potential read > > of uninitialized variable 'virtual_memory' > > Upstream-Status: Pending [merge of multithreading patches to upstream] > > This does look like a good fix. Are these changes to the patch from > upstream? > > Once upstream has accepted the change we should change the status from > "pending", but for now this is ok. > > > 2) CVE-2021-3421.patch > > Removed RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH as > > it is not needed during backporting of original patch. > > Upstream-Status: Backport [ > https://github.com/rpm-software-management/rpm/commit/d6a86b5e69e46cc283b1e06c92343319beb42e21 > <https://apc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Frpm-software-management%2Frpm%2Fcommit%2Fd6a86b5e69e46cc283b1e06c92343319beb42e21=04%7C01%7Cranjitsinh.rathod%40kpit.com%7Cdfd54731b1a240ea64ed08d9785a7618%7C3539451eb46e4a26a242ff61502855c7%7C0%7C0%7C637673152237746428%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000=BFoFI3j9RjhqXQi1tSqfoVoS2strOChMcswosTH59Fs%3D=0> > ] > > Removing these unused definitions doesn't really seem like a critical > issue. I'd prefer to leave the CVE patch in its original form. > > Could you submit a V2 with this change? > > Thanks! > > Steve > > > Signed-off-by: Ranjitsinh Rathod > > --- > > ...rict-virtual-memory-usage-if-limit-s.patch | 25 --- > > .../rpm/files/CVE-2021-3421.patch | 32 +++ > > 2 files changed, 19 insertions(+), 38 deletions(-) > > > > diff --git > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > index 6454785254..dc3f74fecd 100644 > > --- > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > +++ > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > @@ -11,36 +11,39 @@ CPU thread. > > Upstream-Status: Pending [merge of multithreading patches to upstream] >
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
Hi Steve, If you wanted to take changes only for the 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch then you can cherry-pick it from master as I have submitted it for master and it is available on master branch now. Below is the link. poky - Poky Build Tool and Metadata (yoctoproject.org)<https://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=9886ef691aa117d67e4342c6a5e3f79f6a05f8d5> Do you still want me to send v2 patch here? Thanks, Best Regards, Ranjitsinh Rathod Technical Leader | | KPIT Technologies Ltd. Cellphone: +91-84606 92403 __ KPIT<http://www.kpit.com/> | Follow us on LinkedIn<http://www.kpit.com/linkedin> [cid:05fb0115-01bd-4421-ae2f-587c78415386]<https://www.kpit.com/TheNewBrand> From: openembedded-core@lists.openembedded.org on behalf of Alexander Kanavin via lists.openembedded.org Sent: Wednesday, September 15, 2021 8:36 PM To: Steve Sakoman Cc: Ranjitsinh Rathod ; Patches and discussions about the oe-core layer ; Ranjitsinh Rathod Subject: Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code Caution: This email originated from outside of the KPIT. Do not click links or open attachments unless you recognize the sender and know the content is safe. At this point I have to note that I am removing the patch altogether with the upcoming upgrade of rpm to 4.17, as I'm also switching the compression format to zstd, and the patch is generally difficult to maintain and rebase. If you care about xz compression, please do work with upstream to get it merged there. Alex On Wed, 15 Sept 2021 at 16:59, Steve Sakoman mailto:st...@sakoman.com>> wrote: On Wed, Sep 8, 2021 at 4:02 AM Ranjitsinh Rathod mailto:ranjitsinhrathod1...@gmail.com>> wrote: > > From: Ranjitsinh Rathod > mailto:ranjitsinh.rat...@kpit.com>> > > Change in 2 patch as below to avoid critical issues > 1) 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > Handled return values of getrlimit() and lzma_cputhreads() functions > to avoid unexpected behaviours like devide by zero and potential read > of uninitialized variable 'virtual_memory' > Upstream-Status: Pending [merge of multithreading patches to upstream] This does look like a good fix. Are these changes to the patch from upstream? Once upstream has accepted the change we should change the status from "pending", but for now this is ok. > 2) CVE-2021-3421.patch > Removed RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH as > it is not needed during backporting of original patch. > Upstream-Status: Backport > [https://github.com/rpm-software-management/rpm/commit/d6a86b5e69e46cc283b1e06c92343319beb42e21<https://apc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Frpm-software-management%2Frpm%2Fcommit%2Fd6a86b5e69e46cc283b1e06c92343319beb42e21=04%7C01%7Cranjitsinh.rathod%40kpit.com%7Cdfd54731b1a240ea64ed08d9785a7618%7C3539451eb46e4a26a242ff61502855c7%7C0%7C0%7C637673152237746428%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000=BFoFI3j9RjhqXQi1tSqfoVoS2strOChMcswosTH59Fs%3D=0>] Removing these unused definitions doesn't really seem like a critical issue. I'd prefer to leave the CVE patch in its original form. Could you submit a V2 with this change? Thanks! Steve > Signed-off-by: Ranjitsinh Rathod > mailto:ranjitsinh.rat...@kpit.com>> > --- > ...rict-virtual-memory-usage-if-limit-s.patch | 25 --- > .../rpm/files/CVE-2021-3421.patch | 32 +++ > 2 files changed, 19 insertions(+), 38 deletions(-) > > diff --git > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > index 6454785254..dc3f74fecd 100644 > --- > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > +++ > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > @@ -11,36 +11,39 @@ CPU thread. > Upstream-Status: Pending [merge of multithreading patches to upstream] > > Signed-off-by: Peter Bergin > mailto:pe...@berginkonsult.se>> > +Signed-off-by: Ranjitsinh Rathod > mailto:ranjitsinh.rat...@kpit.com>> > --- > - rpmio/rpmio.c | 34 ++ > - 1 file changed, 34 insertions(+) > + rpmio/rpmio.c | 36 > + 1 file changed, 36 insertions(+) > > diff --git a/rpmio/rpmio.c b/rpmio/rpmio.c > index e051c98..b3c56b6 100644 > --- a/rpmio/rpmio.c > +++ b/rpmio/rpmio.c > -@@ -845,6 +845,
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
At this point I have to note that I am removing the patch altogether with the upcoming upgrade of rpm to 4.17, as I'm also switching the compression format to zstd, and the patch is generally difficult to maintain and rebase. If you care about xz compression, please do work with upstream to get it merged there. Alex On Wed, 15 Sept 2021 at 16:59, Steve Sakoman wrote: > On Wed, Sep 8, 2021 at 4:02 AM Ranjitsinh Rathod > wrote: > > > > From: Ranjitsinh Rathod > > > > Change in 2 patch as below to avoid critical issues > > 1) 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > Handled return values of getrlimit() and lzma_cputhreads() functions > > to avoid unexpected behaviours like devide by zero and potential read > > of uninitialized variable 'virtual_memory' > > Upstream-Status: Pending [merge of multithreading patches to upstream] > > This does look like a good fix. Are these changes to the patch from > upstream? > > Once upstream has accepted the change we should change the status from > "pending", but for now this is ok. > > > 2) CVE-2021-3421.patch > > Removed RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH as > > it is not needed during backporting of original patch. > > Upstream-Status: Backport [ > https://github.com/rpm-software-management/rpm/commit/d6a86b5e69e46cc283b1e06c92343319beb42e21 > ] > > Removing these unused definitions doesn't really seem like a critical > issue. I'd prefer to leave the CVE patch in its original form. > > Could you submit a V2 with this change? > > Thanks! > > Steve > > > Signed-off-by: Ranjitsinh Rathod > > --- > > ...rict-virtual-memory-usage-if-limit-s.patch | 25 --- > > .../rpm/files/CVE-2021-3421.patch | 32 +++ > > 2 files changed, 19 insertions(+), 38 deletions(-) > > > > diff --git > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > index 6454785254..dc3f74fecd 100644 > > --- > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > +++ > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > @@ -11,36 +11,39 @@ CPU thread. > > Upstream-Status: Pending [merge of multithreading patches to upstream] > > > > Signed-off-by: Peter Bergin > > +Signed-off-by: Ranjitsinh Rathod > > --- > > - rpmio/rpmio.c | 34 ++ > > - 1 file changed, 34 insertions(+) > > + rpmio/rpmio.c | 36 > > + 1 file changed, 36 insertions(+) > > > > diff --git a/rpmio/rpmio.c b/rpmio/rpmio.c > > index e051c98..b3c56b6 100644 > > --- a/rpmio/rpmio.c > > +++ b/rpmio/rpmio.c > > -@@ -845,6 +845,40 @@ static LZFILE *lzopen_internal(const char *mode, > int fd, int xz) > > +@@ -845,6 +845,42 @@ static LZFILE *lzopen_internal(const char *mode, > int fd, int xz) > > } > > #endif > > > > -+ struct rlimit virtual_memory; > > -+ getrlimit(RLIMIT_AS, _memory); > > -+ if (virtual_memory.rlim_cur != RLIM_INFINITY) { > > ++ struct rlimit virtual_memory = {RLIM_INFINITY , > RLIM_INFINITY}; > > ++ int status = getrlimit(RLIMIT_AS, _memory); > > ++ if ((status != -1) && (virtual_memory.rlim_cur != > RLIM_INFINITY)) { > > + const uint64_t virtual_memlimit = > virtual_memory.rlim_cur; > > ++ uint32_t threads_max = lzma_cputhreads(); > > + const uint64_t virtual_memlimit_per_cpu_thread = > > -+ virtual_memlimit / lzma_cputhreads(); > > -+ uint64_t memory_usage_virt; > > ++ virtual_memlimit / ((threads_max == 0) ? > 1 : threads_max); > > + rpmlog(RPMLOG_NOTICE, "XZ: virtual memory > restricted to %lu and " > > + "per CPU thread %lu\n", virtual_memlimit, > virtual_memlimit_per_cpu_thread); > > ++ uint64_t memory_usage_virt; > > + /* keep reducing the number of compression > threads until memory > > + usage falls below the limit per CPU thread*/ > > + while ((memory_usage_virt = > lzma_stream_encoder_mt_memusage(_options)) > > > + virtual_memlimit_per_cpu_thread) { > > -+ /* If number of threads goes down to > zero lzma_stream_encoder will > > -+ * will return UINT64_MAX. We must check > here to avoid an infinite loop. > > ++ /* If number of threads goes down to > zero or in case of any other error > > ++ * lzma_stream_encoder_mt_memusage will > return UINT64_MAX. We must check > > ++
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
On Wed, Sep 8, 2021 at 4:02 AM Ranjitsinh Rathod wrote: > > From: Ranjitsinh Rathod > > Change in 2 patch as below to avoid critical issues > 1) 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > Handled return values of getrlimit() and lzma_cputhreads() functions > to avoid unexpected behaviours like devide by zero and potential read > of uninitialized variable 'virtual_memory' > Upstream-Status: Pending [merge of multithreading patches to upstream] This does look like a good fix. Are these changes to the patch from upstream? Once upstream has accepted the change we should change the status from "pending", but for now this is ok. > 2) CVE-2021-3421.patch > Removed RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH as > it is not needed during backporting of original patch. > Upstream-Status: Backport > [https://github.com/rpm-software-management/rpm/commit/d6a86b5e69e46cc283b1e06c92343319beb42e21] Removing these unused definitions doesn't really seem like a critical issue. I'd prefer to leave the CVE patch in its original form. Could you submit a V2 with this change? Thanks! Steve > Signed-off-by: Ranjitsinh Rathod > --- > ...rict-virtual-memory-usage-if-limit-s.patch | 25 --- > .../rpm/files/CVE-2021-3421.patch | 32 +++ > 2 files changed, 19 insertions(+), 38 deletions(-) > > diff --git > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > index 6454785254..dc3f74fecd 100644 > --- > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > +++ > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > @@ -11,36 +11,39 @@ CPU thread. > Upstream-Status: Pending [merge of multithreading patches to upstream] > > Signed-off-by: Peter Bergin > +Signed-off-by: Ranjitsinh Rathod > --- > - rpmio/rpmio.c | 34 ++ > - 1 file changed, 34 insertions(+) > + rpmio/rpmio.c | 36 > + 1 file changed, 36 insertions(+) > > diff --git a/rpmio/rpmio.c b/rpmio/rpmio.c > index e051c98..b3c56b6 100644 > --- a/rpmio/rpmio.c > +++ b/rpmio/rpmio.c > -@@ -845,6 +845,40 @@ static LZFILE *lzopen_internal(const char *mode, int > fd, int xz) > +@@ -845,6 +845,42 @@ static LZFILE *lzopen_internal(const char *mode, int > fd, int xz) > } > #endif > > -+ struct rlimit virtual_memory; > -+ getrlimit(RLIMIT_AS, _memory); > -+ if (virtual_memory.rlim_cur != RLIM_INFINITY) { > ++ struct rlimit virtual_memory = {RLIM_INFINITY , > RLIM_INFINITY}; > ++ int status = getrlimit(RLIMIT_AS, _memory); > ++ if ((status != -1) && (virtual_memory.rlim_cur != > RLIM_INFINITY)) { > + const uint64_t virtual_memlimit = > virtual_memory.rlim_cur; > ++ uint32_t threads_max = lzma_cputhreads(); > + const uint64_t virtual_memlimit_per_cpu_thread = > -+ virtual_memlimit / lzma_cputhreads(); > -+ uint64_t memory_usage_virt; > ++ virtual_memlimit / ((threads_max == 0) ? 1 : > threads_max); > + rpmlog(RPMLOG_NOTICE, "XZ: virtual memory restricted > to %lu and " > + "per CPU thread %lu\n", virtual_memlimit, > virtual_memlimit_per_cpu_thread); > ++ uint64_t memory_usage_virt; > + /* keep reducing the number of compression threads > until memory > + usage falls below the limit per CPU thread*/ > + while ((memory_usage_virt = > lzma_stream_encoder_mt_memusage(_options)) > > + virtual_memlimit_per_cpu_thread) { > -+ /* If number of threads goes down to zero > lzma_stream_encoder will > -+ * will return UINT64_MAX. We must check here > to avoid an infinite loop. > ++ /* If number of threads goes down to zero or > in case of any other error > ++ * lzma_stream_encoder_mt_memusage will > return UINT64_MAX. We must check > ++ * for both the cases here to avoid an > infinite loop. > + * If we get into situation that one thread > requires more virtual memory > + * than available we set one thread, print > error message and try anyway. */ > -+ if (--mt_options.threads == 0) { > ++ if ((--mt_options.threads == 0) || > (memory_usage_virt == UINT64_MAX)) { > +
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
[Edited Message Follows] On Mon, Sep 13, 2021 at 11:34 AM, Steve Sakoman wrote: > > RPMSIGTAG_FILESIGNATURELENGTH Hi, Steve and Ranjitsinh, Sorry for the late response. I know that the RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH are defined in the original commit, but are not used. I left it with the author`s intent. If the build goes well without those variables, it doesn't seem to matter. Thanks, Minjae Kim. -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#155997): https://lists.openembedded.org/g/openembedded-core/message/155997 Mute This Topic: https://lists.openembedded.org/mt/85459532/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
On Mon, Sep 13, 2021 at 11:34 AM, Steve Sakoman wrote: > > RPMSIGTAG_FILESIGNATURELENGTH Sorry for the late reponse. I know that the RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH are defined in the original commit, but are not used. I left it with the author`s intent. If the build goes well without those variables, it doesn't seem to matter. -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#155997): https://lists.openembedded.org/g/openembedded-core/message/155997 Mute This Topic: https://lists.openembedded.org/mt/85459532/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
On Wed, Sep 8, 2021 at 4:02 AM Ranjitsinh Rathod wrote: > > From: Ranjitsinh Rathod > > Change in 2 patch as below to avoid critical issues > 1) 0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > Handled return values of getrlimit() and lzma_cputhreads() functions > to avoid unexpected behaviours like devide by zero and potential read > of uninitialized variable 'virtual_memory' > Upstream-Status: Pending [merge of multithreading patches to upstream] > > 2) CVE-2021-3421.patch > Removed RPMSIGTAG_FILESIGNATURES and RPMSIGTAG_FILESIGNATURELENGTH as > it is not needed during backporting of original patch. > Upstream-Status: Backport > [https://github.com/rpm-software-management/rpm/commit/d6a86b5e69e46cc283b1e06c92343319beb42e21] Minjae, can you review this since he is modifying your CVE patch? Thanks! Steve > Signed-off-by: Ranjitsinh Rathod > --- > ...rict-virtual-memory-usage-if-limit-s.patch | 25 --- > .../rpm/files/CVE-2021-3421.patch | 32 +++ > 2 files changed, 19 insertions(+), 38 deletions(-) > > diff --git > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > index 6454785254..dc3f74fecd 100644 > --- > a/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > +++ > b/meta/recipes-devtools/rpm/files/0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch > @@ -11,36 +11,39 @@ CPU thread. > Upstream-Status: Pending [merge of multithreading patches to upstream] > > Signed-off-by: Peter Bergin > +Signed-off-by: Ranjitsinh Rathod > --- > - rpmio/rpmio.c | 34 ++ > - 1 file changed, 34 insertions(+) > + rpmio/rpmio.c | 36 > + 1 file changed, 36 insertions(+) > > diff --git a/rpmio/rpmio.c b/rpmio/rpmio.c > index e051c98..b3c56b6 100644 > --- a/rpmio/rpmio.c > +++ b/rpmio/rpmio.c > -@@ -845,6 +845,40 @@ static LZFILE *lzopen_internal(const char *mode, int > fd, int xz) > +@@ -845,6 +845,42 @@ static LZFILE *lzopen_internal(const char *mode, int > fd, int xz) > } > #endif > > -+ struct rlimit virtual_memory; > -+ getrlimit(RLIMIT_AS, _memory); > -+ if (virtual_memory.rlim_cur != RLIM_INFINITY) { > ++ struct rlimit virtual_memory = {RLIM_INFINITY , > RLIM_INFINITY}; > ++ int status = getrlimit(RLIMIT_AS, _memory); > ++ if ((status != -1) && (virtual_memory.rlim_cur != > RLIM_INFINITY)) { > + const uint64_t virtual_memlimit = > virtual_memory.rlim_cur; > ++ uint32_t threads_max = lzma_cputhreads(); > + const uint64_t virtual_memlimit_per_cpu_thread = > -+ virtual_memlimit / lzma_cputhreads(); > -+ uint64_t memory_usage_virt; > ++ virtual_memlimit / ((threads_max == 0) ? 1 : > threads_max); > + rpmlog(RPMLOG_NOTICE, "XZ: virtual memory restricted > to %lu and " > + "per CPU thread %lu\n", virtual_memlimit, > virtual_memlimit_per_cpu_thread); > ++ uint64_t memory_usage_virt; > + /* keep reducing the number of compression threads > until memory > + usage falls below the limit per CPU thread*/ > + while ((memory_usage_virt = > lzma_stream_encoder_mt_memusage(_options)) > > + virtual_memlimit_per_cpu_thread) { > -+ /* If number of threads goes down to zero > lzma_stream_encoder will > -+ * will return UINT64_MAX. We must check here > to avoid an infinite loop. > ++ /* If number of threads goes down to zero or > in case of any other error > ++ * lzma_stream_encoder_mt_memusage will > return UINT64_MAX. We must check > ++ * for both the cases here to avoid an > infinite loop. > + * If we get into situation that one thread > requires more virtual memory > + * than available we set one thread, print > error message and try anyway. */ > -+ if (--mt_options.threads == 0) { > ++ if ((--mt_options.threads == 0) || > (memory_usage_virt == UINT64_MAX)) { > + mt_options.threads = 1; > + rpmlog(RPMLOG_WARNING, > + "XZ: Could not adjust number > of threads to get below " > diff --git a/meta/recipes-devtools/rpm/files/CVE-2021-3421.patch >
Re: [OE-core] [meta][dunfell][PATCH] rpm: Handle proper return value to avoid major issues and removing unnecessary code
Can someone please check this and confirm if this can go on dunfell? -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#155984): https://lists.openembedded.org/g/openembedded-core/message/155984 Mute This Topic: https://lists.openembedded.org/mt/85459532/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-