| Message ID | ZIpm3pcs3iCP9UaR@work (mailing list archive) |
|---|---|
| State | Mainlined |
| Commit | 36de5f303ca1bd6fce74815ef17ef3d8ff8737b5 |
| Headers | show |
| Series | [next] cgroup: Avoid -Wstringop-overflow warnings | expand |
On Wed, Jun 14, 2023 at 07:18:22PM -0600, Gustavo A. R. Silva wrote: > Address the following -Wstringop-overflow warnings seen when > built with ARM architecture and aspeed_g4_defconfig configuration > (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): > kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > > These changes are based on commit d20d30ebb199 ("cgroup: Avoid compiler > warnings with no subsystems"). > > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Ah nice, this should reduce code size for the "0" case too. Reviewed-by: Kees Cook <keescook@chromium.org>
On Wed, Jun 14, 2023 at 07:18:22PM -0600, Gustavo A. R. Silva wrote: > Address the following -Wstringop-overflow warnings seen when > built with ARM architecture and aspeed_g4_defconfig configuration > (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): > kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > > These changes are based on commit d20d30ebb199 ("cgroup: Avoid compiler > warnings with no subsystems"). > > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Applied to cgroup/for-6.5. Thanks.
On 15/06/2023 02:18, Gustavo A. R. Silva wrote: > Address the following -Wstringop-overflow warnings seen when > built with ARM architecture and aspeed_g4_defconfig configuration > (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): > kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > > These changes are based on commit d20d30ebb199 ("cgroup: Avoid compiler > warnings with no subsystems"). > > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> > --- > kernel/cgroup/cgroup.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c > index cd497b90e11a..1ee76e62eb98 100644 > --- a/kernel/cgroup/cgroup.c > +++ b/kernel/cgroup/cgroup.c > @@ -1200,6 +1200,9 @@ static struct css_set *find_css_set(struct css_set *old_cset, > unsigned long key; > int ssid; > > + if (!CGROUP_HAS_SUBSYS_CONFIG) > + return NULL; > + > lockdep_assert_held(&cgroup_mutex); > > /* First see if we already have a cgroup group that matches > @@ -6045,6 +6048,9 @@ int __init cgroup_init(void) > struct cgroup_subsys *ss; > int ssid; > > + if (!CGROUP_HAS_SUBSYS_CONFIG) > + return -EINVAL; > + > BUILD_BUG_ON(CGROUP_SUBSYS_COUNT > 16); > BUG_ON(cgroup_init_cftypes(NULL, cgroup_base_files)); > BUG_ON(cgroup_init_cftypes(NULL, cgroup_psi_files)); This change (which landed in linux-next) causes a boot failure on my (arm32) board because the cgroup filesystem isn't created which upsets systemd: [ 11.474767] systemd[1]: Failed to mount tmpfs at /sys/fs/cgroup: No such file or directory [ 11.489933] systemd[1]: Failed to mount cgroup at /sys/fs/cgroup/systemd: No such file or directory [!!!!!!] Failed to mount API filesystems. Reverting this commit on the head of linux-next gets the board working again. Thanks, Steve
On 15/06/2023 03:18, Gustavo A. R. Silva wrote: > Address the following -Wstringop-overflow warnings seen when > built with ARM architecture and aspeed_g4_defconfig configuration > (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): > kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > Hi, This patch landed in next but causes boot failures on Exynos ARMv7 board (32 bit) with systemd (exynos_defconfig): NFS-Mount: 192.168.1.10:/srv/nfs/odroidhc1 Waiting 10 seconds for device /dev/nfs ... ERROR: device '/dev/nfs' not found. Skipping fsck. Mount cmd: mount.nfs4 -o vers=4,nolock 192.168.1.10:/srv/nfs/odroidhc1 /new_root :: running cleanup hook [udev] [ 23.752917] systemd[1]: System time before build time, advancing clock. [ 23.851828] systemd[1]: Failed to mount tmpfs at /sys/fs/cgroup: No such file or directory [ 23.868459] systemd[1]: Failed to mount cgroup at /sys/fs/cgroup/systemd: No such file or directory [!!!!!!] Failed to mount API filesystems. [ 23.914562] systemd[1]: Freezing execution. Full log: https://krzk.eu/#/builders/21/builds/4110/steps/15/logs/serial0 Best regards, Krzysztof
Hi! On 6/22/23 04:11, Steven Price wrote: > On 15/06/2023 02:18, Gustavo A. R. Silva wrote: >> Address the following -Wstringop-overflow warnings seen when >> built with ARM architecture and aspeed_g4_defconfig configuration >> (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): >> kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> >> These changes are based on commit d20d30ebb199 ("cgroup: Avoid compiler >> warnings with no subsystems"). >> >> Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> >> --- >> kernel/cgroup/cgroup.c | 6 ++++++ >> 1 file changed, 6 insertions(+) >> >> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c >> index cd497b90e11a..1ee76e62eb98 100644 >> --- a/kernel/cgroup/cgroup.c >> +++ b/kernel/cgroup/cgroup.c >> @@ -1200,6 +1200,9 @@ static struct css_set *find_css_set(struct css_set *old_cset, >> unsigned long key; >> int ssid; >> >> + if (!CGROUP_HAS_SUBSYS_CONFIG) >> + return NULL; >> + >> lockdep_assert_held(&cgroup_mutex); >> >> /* First see if we already have a cgroup group that matches >> @@ -6045,6 +6048,9 @@ int __init cgroup_init(void) >> struct cgroup_subsys *ss; >> int ssid; >> >> + if (!CGROUP_HAS_SUBSYS_CONFIG) >> + return -EINVAL; >> + >> BUILD_BUG_ON(CGROUP_SUBSYS_COUNT > 16); >> BUG_ON(cgroup_init_cftypes(NULL, cgroup_base_files)); >> BUG_ON(cgroup_init_cftypes(NULL, cgroup_psi_files)); > > This change (which landed in linux-next) causes a boot failure on my > (arm32) board because the cgroup filesystem isn't created which upsets > systemd: > > [ 11.474767] systemd[1]: Failed to mount tmpfs at /sys/fs/cgroup: No such file or directory > [ 11.489933] systemd[1]: Failed to mount cgroup at /sys/fs/cgroup/systemd: No such file or directory > [!!!!!!] Failed to mount API filesystems. > > Reverting this commit on the head of linux-next gets the board working > again. Thanks for reporting this issue. I'll take a look. -- Gustavo
Hi! On 6/22/23 04:20, Krzysztof Kozlowski wrote: > On 15/06/2023 03:18, Gustavo A. R. Silva wrote: >> Address the following -Wstringop-overflow warnings seen when >> built with ARM architecture and aspeed_g4_defconfig configuration >> (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): >> kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] >> > > Hi, > > This patch landed in next but causes boot failures on > Exynos ARMv7 board (32 bit) with systemd (exynos_defconfig): > > NFS-Mount: 192.168.1.10:/srv/nfs/odroidhc1 > Waiting 10 seconds for device /dev/nfs ... > ERROR: device '/dev/nfs' not found. Skipping fsck. > Mount cmd: > mount.nfs4 -o vers=4,nolock 192.168.1.10:/srv/nfs/odroidhc1 /new_root > :: running cleanup hook [udev] > [ 23.752917] systemd[1]: System time before build time, advancing clock. > [ 23.851828] systemd[1]: Failed to mount tmpfs at /sys/fs/cgroup: No such file or directory > [ 23.868459] systemd[1]: Failed to mount cgroup at /sys/fs/cgroup/systemd: No such file or directory > [!!!!!!] Failed to mount API filesystems. > [ 23.914562] systemd[1]: Freezing execution. > > Full log: > https://krzk.eu/#/builders/21/builds/4110/steps/15/logs/serial0 Thanks for reporting this issue. I'll take a look. -- Gustavo
On Thu, Jun 22, 2023 at 08:09:03AM -0600, Gustavo A. R. Silva wrote: > > Reverting this commit on the head of linux-next gets the board working > > again. > > Thanks for reporting this issue. I'll take a look. I'm reverting the commit from cgroup/for-6.5. Let's try again later. Thanks.
Hi all, I wonder if you have any suggestions on how to address this issue. As it seems that my last attempt caused some boot failures[1][2]. At first, I thought that the right way to fix this was through a similar fix as this one[3]. But it seems I'm missing something else that I cannot determine yet. These -Wstringop-overflow warnings are mostly the last ones remaining before we can finally enable this compiler option, globally. Any help or advice on how to properly address this is greatly appreciated. :) Thanks! -- Gustavo [1] https://lore.kernel.org/linux-hardening/726aae97-755d-9806-11d4-2fb21aa93428@arm.com/ [2] https://lore.kernel.org/linux-hardening/361c2f87-1424-f452-912f-0e4a339f5c46@kernel.org/ [3] https://git.kernel.org/linus/d20d30ebb199 On 6/14/23 19:18, Gustavo A. R. Silva wrote: > Address the following -Wstringop-overflow warnings seen when > built with ARM architecture and aspeed_g4_defconfig configuration > (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): > kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] > > These changes are based on commit d20d30ebb199 ("cgroup: Avoid compiler > warnings with no subsystems"). > > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> > --- > kernel/cgroup/cgroup.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c > index cd497b90e11a..1ee76e62eb98 100644 > --- a/kernel/cgroup/cgroup.c > +++ b/kernel/cgroup/cgroup.c > @@ -1200,6 +1200,9 @@ static struct css_set *find_css_set(struct css_set *old_cset, > unsigned long key; > int ssid; > > + if (!CGROUP_HAS_SUBSYS_CONFIG) > + return NULL; > + > lockdep_assert_held(&cgroup_mutex); > > /* First see if we already have a cgroup group that matches > @@ -6045,6 +6048,9 @@ int __init cgroup_init(void) > struct cgroup_subsys *ss; > int ssid; > > + if (!CGROUP_HAS_SUBSYS_CONFIG) > + return -EINVAL; > + > BUILD_BUG_ON(CGROUP_SUBSYS_COUNT > 16); > BUG_ON(cgroup_init_cftypes(NULL, cgroup_base_files)); > BUG_ON(cgroup_init_cftypes(NULL, cgroup_psi_files));
On Tue, Aug 15, 2023 at 01:04:32PM -0600, Gustavo A. R. Silva wrote: > Hi all, > > I wonder if you have any suggestions on how to address this issue. As it seems that > my last attempt caused some boot failures[1][2]. > > At first, I thought that the right way to fix this was through a similar fix as this > one[3]. But it seems I'm missing something else that I cannot determine yet. > > These -Wstringop-overflow warnings are mostly the last ones remaining before we can > finally enable this compiler option, globally. > > Any help or advice on how to properly address this is greatly appreciated. :) It looks like there are only two functions which misbehave when there are no controllers configured. Maybe just fix them so that they don't access empty array? Thanks.
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index cd497b90e11a..1ee76e62eb98 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -1200,6 +1200,9 @@ static struct css_set *find_css_set(struct css_set *old_cset, unsigned long key; int ssid; + if (!CGROUP_HAS_SUBSYS_CONFIG) + return NULL; + lockdep_assert_held(&cgroup_mutex); /* First see if we already have a cgroup group that matches @@ -6045,6 +6048,9 @@ int __init cgroup_init(void) struct cgroup_subsys *ss; int ssid; + if (!CGROUP_HAS_SUBSYS_CONFIG) + return -EINVAL; + BUILD_BUG_ON(CGROUP_SUBSYS_COUNT > 16); BUG_ON(cgroup_init_cftypes(NULL, cgroup_base_files)); BUG_ON(cgroup_init_cftypes(NULL, cgroup_psi_files));
Address the following -Wstringop-overflow warnings seen when built with ARM architecture and aspeed_g4_defconfig configuration (notice that under this configuration CGROUP_SUBSYS_COUNT == 0): kernel/cgroup/cgroup.c:1208:16: warning: 'find_existing_css_set' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] kernel/cgroup/cgroup.c:1258:15: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] kernel/cgroup/cgroup.c:6089:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] kernel/cgroup/cgroup.c:6153:18: warning: 'css_set_hash' accessing 4 bytes in a region of size 0 [-Wstringop-overflow=] These changes are based on commit d20d30ebb199 ("cgroup: Avoid compiler warnings with no subsystems"). Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> --- kernel/cgroup/cgroup.c | 6 ++++++ 1 file changed, 6 insertions(+)