replace runtime slub_debug toggling with more capable boot parameter
Subsystems (3)
cover replace runtime slub_debug toggling with more capable boot parameter
Hi,
we had recently reports [1,2] of troubles with runtime writes to various SLUB per-cache sysfs files related to slub_debug or tuning leading to crashes. I have inspected all those writable files and the rather unfortunate result is that most of them are made read-only by this patchset, as fixing the issues doesn't seem viable. Details for each are given in each patch. I expect some resistance, hence the RFC.
The runtime toggles were however necessary for the use case described in [3], so the first patch extends the slub_debug boot parameter syntax to achieve the same configuration without runtime toggles. That should hopefully make the changes more feasible.
[1] https://lkml.kernel.org/r/[email protected]
[2] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
[3] https://lore.kernel.org/r/[email protected]
Vlastimil Babka (5):
mm, slub: extend slub_debug syntax for multiple blocks mm, slub: make some slub_debug related attributes read-only mm, slub: remove runtime allocation order changes
mm, slub: make remaining slub_debug related attributes read-only mm, slub: make reclaim_account attribute read-only
.../admin-guide/kernel-parameters.txt | 2 +- Documentation/vm/slub.rst | 36 +- mm/slub.c | 315 ++++++++---------- 3 files changed, 163 insertions(+), 190 deletions(-)
Signature · 1 line
2.26.2
RFC PATCH 1/5 mm, slub: extend slub_debug syntax for multiple blocks
The slub_debug kernel boot parameter can either apply a single set of options to all caches or a list of caches. There is a use case where debugging is applied for all caches and then disabled at runtime for specific caches, for performance and memory consumption reasons [1]. As runtime changes are dangerous, extend the boot parameter syntax so that multiple blocks of either global or slab-specific options can be specified, with blocks delimited by ';'. This will also support the use case of [1] without runtime changes.
For details see the updated Documentation/vm/slub.rst
[1] https://lore.kernel.org/r/[email protected]
--- .../admin-guide/kernel-parameters.txt | 2 +- Documentation/vm/slub.rst | 17 ++ mm/slub.c | 177 +++++++++++++----- 3 files changed, 144 insertions(+), 52 deletions(-)
Documentation/admin-guide/kernel-parameters.txt +1 -1
@@ -4613,7 +4613,7 @@ fragmentation. Defaults to 1 for systems with more than 32MB of RAM, 0 otherwise. - slub_debug[=options[,slabs]] [MM, SLUB] + slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB] Enabling slub_debug allows one to determine the culprit if slab objects become corrupted. Enabling slub_debug can create guard zones around objects and
Documentation/vm/slub.rst +17 -0
@@ -41,6 +41,11 @@ slub_debug=<Debug-Options>,<slab name1>,<slab name2>,... Enable options only for select slabs (no spaces after a comma) +Multiple blocks of options for all slabs or selected slabs can be given, with +blocks of options delimited by ';'. The last of "all slabs" blocks is applied +to all slabs except those that match one of the "select slabs" block. Options +of the first "select slabs" blocks that matches the slab's name are applied. + Possible debug options are:: F Sanity checks on (enables SLAB_DEBUG_CONSISTENCY_CHECKS
@@ -83,6 +88,18 @@ in low memory situations or if there's high fragmentation of memory. To slub_debug=O +You can apply different options to different list of slab names, using blocks +of options. This will enable red zoning for dentry and user tracking for +kmalloc. All other slabs will not get any debugging enabled:: + + slub_debug=Z,dentry;U,kmalloc-* + +You can also enable options (e.g. sanity checks and poisoning) for all caches +except some that are deemed too performance critical and don't need to be +debugged:: + + slub_debug=FZ;-,zs_handle,zspage + In case you forgot to enable debugging on the kernel command line: It is possible to enable debugging manually when the kernel is up. Look at the contents of::
mm/slub.c +126 -51
237 lines not on this page · raw message
Signature · 1 line
2.26.2
Re: [RFC PATCH 1/5] mm, slub: extend slub_debug syntax for multiple blocks
Quoted 51 lines · Vlastimil Babka
The
slub_debugkernel boot parameter can either apply a single set of options to all caches or a list of caches. There is a use case where debugging is applied for all caches and then disabled at runtime for specific caches, for performance and memory consumption reasons [1]. As runtime changes are dangerous, extend the boot parameter syntax so that multiple blocks of either global or slab-specific options can be specified, with blocks delimited by ';'. This will also support the use case of [1] without runtime changes.For details see the updated
Documentation/vm/slub.rst[1] https://lore.kernel.org/r/[email protected]
Signed-off-by Vlastimil Babka <[email protected]>--- .../admin-guide/kernel-parameters.txt | 2 +- Documentation/vm/slub.rst | 17 ++ mm/slub.c | 177 +++++++++++++----- 3 files changed, 144 insertions(+), 52 deletions(-)Documentation/admin-guide/kernel-parameters.txt +1
-1@@ -4613,7 +4613,7 @@ fragmentation. Defaults to 1 for systems with more than 32MB of RAM, 0 otherwise. - slub_debug[=options[,slabs]] [MM, SLUB] + slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB] Enabling slub_debug allows one to determine the culprit if slab objects become corrupted. Enabling slub_debug can create guard zones around objects andDocumentation/vm/slub.rst +5
-0@@ -41,6 +41,11 @@ slub_debug=<Debug-Options>,<slab name1>,<slab name2>,... Enable options only for select slabs (no spaces after a comma) +Multiple blocks of options for all slabs or selected slabs can be given, with +blocks of options delimited by ';'. The last of "all slabs" blocks is applied +to all slabs except those that match one of the "select slabs" block. Options +of the first "select slabs" blocks that matches the slab's name are applied. + Possible debug options are:: F Sanity checks on (enables SLAB_DEBUG_CONSISTENCY_CHECKS@@ -83,6 +88,18 @@ in low memory situations or if there's high fragmentation of memory. To slub_debug=O
Documentation/vm/slub.rst in low memory situations or if there's high fragmentation of memory. To · line 99
Quoted 10 lines
Documentation/vm/slub.rst +9
-0+You can apply different options to different list of slab names, using blocks +of options. This will enable red zoning for dentry and user tracking for +kmalloc. All other slabs will not get any debugging enabled:: + + slub_debug=Z,dentry;U,kmalloc-* + +You can also enable options (e.g. sanity checks and poisoning) for all caches +except some that are deemed too performance critical and don't need to be +debugged::
Just for more clarity, how about:
... debugged by starting the list with "-" (to mean "all except the
following")::
Documentation/vm/slub.rst in low memory situations or if there's high fragmentation of memory. To · line 105
Quoted 6 lines
Documentation/vm/slub.rst +3
-0+ + slub_debug=FZ;-,zs_handle,zspage + In case you forgot to enable debugging on the kernel command line: It is possible to enable debugging manually when the kernel is up. Look at the contents of::
Everything else looks great; very nice! :)
Signature · 1 line
Kees Cook
Re: [RFC PATCH 1/5] mm, slub: extend slub_debug syntax for multiple blocks
Quoted 16 lines · Kees Cook
Documentation/vm/slub.rst in low memory situations or if there's high fragmentation of memory. To · line 99
Quoted 13 lines
Documentation/vm/slub.rst +9
-0@@ -83,6 +88,18 @@ in low memory situations or if there's high fragmentation of memory. To slub_debug=O +You can apply different options to different list of slab names, using blocks +of options. This will enable red zoning for dentry and user tracking for +kmalloc. All other slabs will not get any debugging enabled:: + + slub_debug=Z,dentry;U,kmalloc-* + +You can also enable options (e.g. sanity checks and poisoning) for all caches +except some that are deemed too performance critical and don't need to be +debugged::Just for more clarity, how about:
Quoted 2 lines
... debugged by starting the list with "-" (to mean "all except the
following")::
Hmm, "-" is not exactly "all except the following", but "no debugging", as explained in the list of debug options earlier in the file.
So I'm updating it to this:
You can also enable options (e.g. sanity checks and poisoning) for all caches except some that are deemed too performance critical and don't need to be debugged by specifying global debug options followed by a list of slab names with "-" as options::
slub_debug=FZ;-,zs_handle,zspage
Quoted 10 lines
Documentation/vm/slub.rst in low memory situations or if there's high fragmentation of memory. To · line 105
Quoted 6 lines
Documentation/vm/slub.rst +3
-0+ + slub_debug=FZ;-,zs_handle,zspage + In case you forgot to enable debugging on the kernel command line: It is possible to enable debugging manually when the kernel is up. Look at the contents of::Everything else looks great; very nice! :)
Reviewed-by Kees Cook <[email protected]>
Thanks!
RFC PATCH 2/5 mm, slub: make some slub_debug related attributes read-only
In reply to Vlastimil BabkaSLUB_DEBUG creates several files under /sys/kernel/slab/<cache>/ that can be read to check if the respective debugging options are enabled for given cache. The options can be also toggled at runtime by writing into the files. Some of those, namely red_zone, poison, and store_user can be toggled only when no objects yet exist in the cache.
Vijayanand reports [1] that there is a problem with freelist randomization if changing the debugging option's state results in different number of objects per page, and the random sequence cache needs thus needs to be recomputed.
However, another problem is that the check for "no objects yet exist in the cache" is racy, as noted by Jann [2] and fixing that would add overhead or otherwise complicate the allocation/freeing paths. Thus it would be much simpler just to remove the runtime toggling support. The documentation describes it's "In case you forgot to enable debugging on the kernel command line", but the neccessity of having no objects limits its usefulness anyway for many caches.
Vijayanand describes an use case [3] where debugging is enabled for all but zram caches for memory overhead reasons, and using the runtime toggles was the only way to achieve such configuration. After the previous patch it's now possible to do that directly from the kernel boot option, so we can remove the dangerous runtime toggles by making the /sys attribute files read-only.
While updating it, also improve the documentation of the debugging /sys files.
[1] https://lkml.kernel.org/r/[email protected]
[2] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
[3] https://lore.kernel.org/r/[email protected]
--- Documentation/vm/slub.rst | 28 ++++++++++++++---------- mm/slub.c | 46 +++------------------------------------ 2 files changed, 20 insertions(+), 54 deletions(-)
Documentation/vm/slub.rst +17 -11
@@ -100,20 +100,26 @@ except some that are deemed too performance critical and don't need to be slub_debug=FZ;-,zs_handle,zspage -In case you forgot to enable debugging on the kernel command line: It is -possible to enable debugging manually when the kernel is up. Look at the -contents of:: +The state of each debug option for a slab can be found in the respective files +under:: /sys/kernel/slab/<slab name>/ -Look at the writable files. Writing 1 to them will enable the -corresponding debug option. All options can be set on a slab that does -not contain objects. If the slab already contains objects then sanity checks -and tracing may only be enabled. The other options may cause the realignment -of objects. - -Careful with tracing: It may spew out lots of information and never stop if -used on the wrong slab. +If the file contains 1, the option is enabled, 0 means disabled. The debug +options from the ``slub_debug`` parameter translate to the following files:: + + F sanity_checks + Z red_zone + P poison + U store_user + T trace + A failslab + +The sanity_checks, trace and failslab files are writable, so writing 1 or 0 +will enable or disable the option at runtime. The writes to trace and failslab +may return -EINVAL if the cache is subject to slab merging. Careful with +tracing: It may spew out lots of information and never stop if used on the +wrong slab. Slab merging ============
mm/slub.c +3 -43
64 lines not on this page · raw message
Signature · 1 line
2.26.2
Re: [RFC PATCH 2/5] mm, slub: make some slub_debug related attributes read-only
Quoted 26 lines · Vlastimil Babka
SLUB_DEBUGcreates several files under /sys/kernel/slab/<cache>/ that can be read to check if the respective debugging options are enabled for given cache. The options can be also toggled at runtime by writing into the files. Some of those, namelyred_zone, poison, andstore_usercan be toggled only when no objects yet exist in the cache.Vijayanand reports [1] that there is a problem with freelist randomization if changing the debugging option's state results in different number of objects per page, and the random sequence cache needs thus needs to be recomputed.
However, another problem is that the check for "no objects yet exist in the cache" is racy, as noted by Jann [2] and fixing that would add overhead or otherwise complicate the allocation/freeing paths. Thus it would be much simpler just to remove the runtime toggling support. The documentation describes it's "In case you forgot to enable debugging on the kernel command line", but the neccessity of having no objects limits its usefulness anyway for many caches.
Vijayanand describes an use case [3] where debugging is enabled for all but zram caches for memory overhead reasons, and using the runtime toggles was the only way to achieve such configuration. After the previous patch it's now possible to do that directly from the kernel boot option, so we can remove the dangerous runtime toggles by making the /sys attribute files read-only.
While updating it, also improve the documentation of the debugging /sys files.
Quoted 7 lines
[1] https://lkml.kernel.org/r/[email protected]
[2] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
[3] https://lore.kernel.org/r/[email protected]Reported-by Vijayanand Jitta <[email protected]>Reported-by Jann Horn <[email protected]>Signed-off-by Vlastimil Babka <[email protected]>
Signature · 1 line
Kees Cook
Re: [RFC PATCH 2/5] mm, slub: make some slub_debug related attributes read-only
In reply to Vlastimil BabkaQuoted 26 lines · Vlastimil Babka
SLUB_DEBUGcreates several files under /sys/kernel/slab/<cache>/ that can be read to check if the respective debugging options are enabled for given cache. The options can be also toggled at runtime by writing into the files. Some of those, namelyred_zone, poison, andstore_usercan be toggled only when no objects yet exist in the cache.Vijayanand reports [1] that there is a problem with freelist randomization if changing the debugging option's state results in different number of objects per page, and the random sequence cache needs thus needs to be recomputed.
However, another problem is that the check for "no objects yet exist in the cache" is racy, as noted by Jann [2] and fixing that would add overhead or otherwise complicate the allocation/freeing paths. Thus it would be much simpler just to remove the runtime toggling support. The documentation describes it's "In case you forgot to enable debugging on the kernel command line", but the neccessity of having no objects limits its usefulness anyway for many caches.
Vijayanand describes an use case [3] where debugging is enabled for all but zram caches for memory overhead reasons, and using the runtime toggles was the only way to achieve such configuration. After the previous patch it's now possible to do that directly from the kernel boot option, so we can remove the dangerous runtime toggles by making the /sys attribute files read-only.
While updating it, also improve the documentation of the debugging /sys files.
Quoted 7 lines
[1] https://lkml.kernel.org/r/[email protected]
[2] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
[3] https://lore.kernel.org/r/[email protected]Reported-by Vijayanand Jitta <[email protected]>Reported-by Jann Horn <[email protected]>Signed-off-by Vlastimil Babka <[email protected]>
Thanks!
Quoted 120 lines
--- Documentation/vm/slub.rst | 28 ++++++++++++++---------- mm/slub.c | 46 +++------------------------------------ 2 files changed, 20 insertions(+), 54 deletions(-)Documentation/vm/slub.rst +17
-11@@ -100,20 +100,26 @@ except some that are deemed too performance critical and don't need to be slub_debug=FZ;-,zs_handle,zspage -In case you forgot to enable debugging on the kernel command line: It is -possible to enable debugging manually when the kernel is up. Look at the -contents of:: +The state of each debug option for a slab can be found in the respective files +under:: /sys/kernel/slab/<slab name>/ -Look at the writable files. Writing 1 to them will enable the -corresponding debug option. All options can be set on a slab that does -not contain objects. If the slab already contains objects then sanity checks -and tracing may only be enabled. The other options may cause the realignment -of objects. - -Careful with tracing: It may spew out lots of information and never stop if -used on the wrong slab. +If the file contains 1, the option is enabled, 0 means disabled. The debug +options from the ``slub_debug`` parameter translate to the following files:: + + F sanity_checks + Z red_zone + P poison + U store_user + T trace + A failslab + +The sanity_checks, trace and failslab files are writable, so writing 1 or 0 +will enable or disable the option at runtime. The writes to trace and failslab +may return -EINVAL if the cache is subject to slab merging. Careful with +tracing: It may spew out lots of information and never stop if used on the +wrong slab. Slab merging ============mm/slub.c +3
-43@@ -5351,61 +5351,21 @@ static ssize_t red_zone_show(struct kmem_cache *s, char *buf)return sprintf (buf,"%d\n" , !!(s->flags &SLAB_RED_ZONE )); } -static ssize_t red_zone_store (struct kmem_cache *s, -const char *buf,size_t length) -{ -if (any_slab_objects (s)) -return -EBUSY ; - - s->flags &= ~SLAB_RED_ZONE ; -if (buf[0 ] =='1' ) { - s->flags |=SLAB_RED_ZONE ; - } -calculate_sizes (s,-1 ); -return length; -} -SLAB_ATTR (red_zone); +SLAB_ATTR_RO (red_zone);static ssize_t poison_show (struct kmem_cache *s,char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_POISON )); } -static ssize_t poison_store (struct kmem_cache *s, -const char *buf,size_t length) -{ -if (any_slab_objects (s)) -return -EBUSY ; - - s->flags &= ~SLAB_POISON ; -if (buf[0 ] =='1' ) { - s->flags |=SLAB_POISON ; - } -calculate_sizes (s,-1 ); -return length; -} -SLAB_ATTR (poison); +SLAB_ATTR_RO (poison);static ssize_t store_user_show (struct kmem_cache *s,char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_STORE_USER )); } -static ssize_t store_user_store (struct kmem_cache *s, -const char *buf,size_t length) -{ -if (any_slab_objects (s)) -return -EBUSY ; - - s->flags &= ~SLAB_STORE_USER ; -if (buf[0 ] =='1' ) { - s->flags &= ~__CMPXCHG_DOUBLE; - s->flags |=SLAB_STORE_USER ; - } -calculate_sizes (s,-1 ); -return length; -} -SLAB_ATTR (store_user); +SLAB_ATTR_RO (store_user);static ssize_t validate_show (struct kmem_cache *s,char *buf) {Signature · 1 line
2.26.2
RFC PATCH 3/5 mm, slub: remove runtime allocation order changes
In reply to Vlastimil BabkaSLUB allows runtime changing of page allocation order by writing into the /sys/kernel/slab/<cache>/order file. Jann has reported [1] that this interface allows the order to be set too small, leading to crashes.
While it's possible to fix the immediate issue, closer inspection reveals potential races. Storing the new order calls calculate_sizes() which non-atomically updates a lot of kmem_cache fields while the cache is still in use. Unexpected behavior might occur even if the fields are set to the same value as they were.
This could be fixed by splitting out the part of calculate_sizes() that depends on forced_order, so that we only update kmem_cache.oo field. This could still race with init_cache_random_seq(), shuffle_freelist(), allocate_slab(). Perhaps it's possible to audit and e.g. add some READ_ONCE/WRITE_ONCE accesses, it might be easier just to remove the runtime order changes, which is what this patch does. If there are valid usecases for per-cache order setting, we could e.g. extend the boot parameters to do that.
[1] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
--- mm/slub.c | 19 +------------------ 1 file changed, 1 insertion(+), 18 deletions(-)
mm/slub.c +1 -18
@@ -5111,28 +5111,11 @@ static ssize_t objs_per_slab_show(struct kmem_cache *s, char *buf) }SLAB_ATTR_RO (objs_per_slab); -static ssize_t order_store (struct kmem_cache *s, -const char *buf,size_t length) -{ -unsigned int order; -int err; - - err =kstrtouint (buf,10 , &order); -if (err) -return err; - -if (order > slub_max_order || order < slub_min_order) -return -EINVAL ; - -calculate_sizes (s, order); -return length; -} -static ssize_t order_show (struct kmem_cache *s,char *buf) {return sprintf (buf,"%u\n" ,oo_order (s->oo)); } -SLAB_ATTR (order); +SLAB_ATTR_RO (order);static ssize_t min_partial_show (struct kmem_cache *s,char *buf) {
Signature · 1 line
2.26.2
Re: [RFC PATCH 3/5] mm, slub: remove runtime allocation order changes
Quoted 18 lines · Vlastimil Babka
SLUB allows runtime changing of page allocation order by writing into the /sys/kernel/slab/<cache>/order file. Jann has reported [1] that this interface allows the order to be set too small, leading to crashes.
While it's possible to fix the immediate issue, closer inspection reveals potential races. Storing the new order calls
calculate_sizes()which non-atomically updates a lot ofkmem_cachefields while the cache is still in use. Unexpected behavior might occur even if the fields are set to the same value as they were.This could be fixed by splitting out the part of
calculate_sizes()that depends onforced_order, so that we only update kmem_cache.oo field. This could still race withinit_cache_random_seq(),shuffle_freelist(),allocate_slab(). Perhaps it's possible to audit and e.g. add some READ_ONCE/WRITE_ONCE accesses, it might be easier just to remove the runtime order changes, which is what this patch does. If there are valid usecases for per-cache order setting, we could e.g. extend the boot parameters to do that.
Quoted 4 lines
[1] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
Reported-by Jann Horn <[email protected]>Signed-off-by Vlastimil Babka <[email protected]>
Signature · 1 line
Kees Cook
Re: [RFC PATCH 3/5] mm, slub: remove runtime allocation order changes
In reply to Vlastimil BabkaQuoted 18 lines · Vlastimil Babka
SLUB allows runtime changing of page allocation order by writing into the /sys/kernel/slab/<cache>/order file. Jann has reported [1] that this interface allows the order to be set too small, leading to crashes.
While it's possible to fix the immediate issue, closer inspection reveals potential races. Storing the new order calls
calculate_sizes()which non-atomically updates a lot ofkmem_cachefields while the cache is still in use. Unexpected behavior might occur even if the fields are set to the same value as they were.This could be fixed by splitting out the part of
calculate_sizes()that depends onforced_order, so that we only update kmem_cache.oo field. This could still race withinit_cache_random_seq(),shuffle_freelist(),allocate_slab(). Perhaps it's possible to audit and e.g. add some READ_ONCE/WRITE_ONCE accesses, it might be easier just to remove the runtime order changes, which is what this patch does. If there are valid usecases for per-cache order setting, we could e.g. extend the boot parameters to do that.
Quoted 4 lines
[1] https://lore.kernel.org/r/CAG48ez31PP--h6_FzVyfJ4H86QYczAFPdxtJHUEEan+7VJETAQ@mail.gmail.com
Reported-by Jann Horn <[email protected]>Signed-off-by Vlastimil Babka <[email protected]>
Thanks!
Quoted 42 lines
--- mm/slub.c | 19 +------------------ 1 file changed, 1 insertion(+), 18 deletions(-)mm/slub.c +1
-18@@ -5111,28 +5111,11 @@ static ssize_t objs_per_slab_show(struct kmem_cache *s, char *buf) }SLAB_ATTR_RO (objs_per_slab); -static ssize_t order_store (struct kmem_cache *s, -const char *buf,size_t length) -{ -unsigned int order; -int err; - - err =kstrtouint (buf,10 , &order); -if (err) -return err; - -if (order > slub_max_order || order < slub_min_order) -return -EINVAL ; - -calculate_sizes (s, order); -return length; -} -static ssize_t order_show (struct kmem_cache *s,char *buf) {return sprintf (buf,"%u\n" ,oo_order (s->oo)); } -SLAB_ATTR (order); +SLAB_ATTR_RO (order);static ssize_t min_partial_show (struct kmem_cache *s,char *buf) {Signature · 1 line
2.26.2
RFC PATCH 4/5 mm, slub: make remaining slub_debug related attributes read-only
In reply to Vlastimil BabkaSLUB_DEBUG creates several files under /sys/kernel/slab/<cache>/ that can be read to check if the respective debugging options are enabled for given cache. Some options, namely sanity_checks, trace, and failslab can be also enabled and disabled at runtime by writing into the files.
The runtime toggling is racy. Some options disable __CMPXCHG_DOUBLE when enabled, which means that in case of concurrent allocations, some can still use __CMPXCHG_DOUBLE and some not, leading to potential corruption. The s->flags field is also not updated or checked atomically. The simplest solution is to remove the runtime toggling. The extended slub_debug boot parameter syntax introduced by earlier patch should allow to fine-tune the debugging configuration during boot with same granularity.
--- Documentation/vm/slub.rst | 7 ++--- mm/slub.c | 62 ++------------------------------------- 2 files changed, 5 insertions(+), 64 deletions(-)
Documentation/vm/slub.rst +2 -5
@@ -115,11 +115,8 @@ If the file contains 1, the option is enabled, 0 means disabled. The debug T trace A failslab -The sanity_checks, trace and failslab files are writable, so writing 1 or 0 -will enable or disable the option at runtime. The writes to trace and failslab -may return -EINVAL if the cache is subject to slab merging. Careful with -tracing: It may spew out lots of information and never stop if used on the -wrong slab. +Careful with tracing: It may spew out lots of information and never stop if +used on the wrong slab. Slab merging ============
mm/slub.c +3 -59
85 lines not on this page · raw message
Signature · 1 line
2.26.2
Re: [RFC PATCH 4/5] mm, slub: make remaining slub_debug related attributes read-only
Quoted 13 lines · Vlastimil Babka
SLUB_DEBUGcreates several files under /sys/kernel/slab/<cache>/ that can be read to check if the respective debugging options are enabled for given cache. Some options, namelysanity_checks, trace, and failslab can be also enabled and disabled at runtime by writing into the files.The runtime toggling is racy. Some options disable
__CMPXCHG_DOUBLEwhen enabled, which means that in case of concurrent allocations, some can still use__CMPXCHG_DOUBLEand some not, leading to potential corruption. Thes->flagsfield is also not updated or checked atomically. The simplest solution is to remove the runtime toggling. The extendedslub_debugboot parameter syntax introduced by earlier patch should allow to fine-tune the debugging configuration during boot with same granularity.
Quoted 1 line
Signed-off-by Vlastimil Babka <[email protected]>
Signature · 1 line
Kees Cook
Re: [RFC PATCH 4/5] mm, slub: make remaining slub_debug related attributes read-only
In reply to Vlastimil BabkaQuoted 13 lines · Vlastimil Babka
SLUB_DEBUGcreates several files under /sys/kernel/slab/<cache>/ that can be read to check if the respective debugging options are enabled for given cache. Some options, namelysanity_checks, trace, and failslab can be also enabled and disabled at runtime by writing into the files.The runtime toggling is racy. Some options disable
__CMPXCHG_DOUBLEwhen enabled, which means that in case of concurrent allocations, some can still use__CMPXCHG_DOUBLEand some not, leading to potential corruption. Thes->flagsfield is also not updated or checked atomically. The simplest solution is to remove the runtime toggling. The extendedslub_debugboot parameter syntax introduced by earlier patch should allow to fine-tune the debugging configuration during boot with same granularity.
Quoted 1 line
Signed-off-by Vlastimil Babka <[email protected]>
Thanks!
Quoted 119 lines
--- Documentation/vm/slub.rst | 7 ++--- mm/slub.c | 62 ++------------------------------------- 2 files changed, 5 insertions(+), 64 deletions(-)Documentation/vm/slub.rst +2
-5@@ -115,11 +115,8 @@ If the file contains 1, the option is enabled, 0 means disabled. The debug T trace A failslab -The sanity_checks, trace and failslab files are writable, so writing 1 or 0 -will enable or disable the option at runtime. The writes to trace and failslab -may return -EINVAL if the cache is subject to slab merging. Careful with -tracing: It may spew out lots of information and never stop if used on the -wrong slab. +Careful with tracing: It may spew out lots of information and never stop if +used on the wrong slab. Slab merging ============mm/slub.c +3
-59@@ -5056,20 +5056,6 @@ static ssize_t show_slab_objects(struct kmem_cache *s,return x +sprintf (buf + x,"\n" ); } -#ifdef CONFIG_SLUB_DEBUG -static int any_slab_objects (struct kmem_cache *s) -{ -int node; -struct kmem_cache_node *n; - -for_each_kmem_cache_node (s, node, n) -if (atomic_long_read (&n->total_objects)) -return 1 ; - -return 0 ; -} -#endif -#define to_slab_attr (n) container_of(n, struct slab_attribute, attr)#define to_slab (n) container_of(n, struct kmem_cache, kobj)@@ -5291,43 +5277,13 @@ static ssize_t sanity_checks_show(struct kmem_cache *s, char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_CONSISTENCY_CHECKS )); } - -static ssize_t sanity_checks_store (struct kmem_cache *s, -const char *buf,size_t length) -{ - s->flags &= ~SLAB_CONSISTENCY_CHECKS ; -if (buf[0 ] =='1' ) { - s->flags &= ~__CMPXCHG_DOUBLE; - s->flags |=SLAB_CONSISTENCY_CHECKS ; - } -return length; -} -SLAB_ATTR (sanity_checks); +SLAB_ATTR_RO (sanity_checks);static ssize_t trace_show (struct kmem_cache *s,char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_TRACE )); } - -static ssize_t trace_store (struct kmem_cache *s,const char *buf, -size_t length) -{ -/* -* Tracing a merged cache is going to give confusing results -* as well as cause other issues like converting a mergeable -* cache into an umergeable one. -*/ -if (s->refcount >1 ) -return -EINVAL ; - - s->flags &= ~SLAB_TRACE ; -if (buf[0 ] =='1' ) { - s->flags &= ~__CMPXCHG_DOUBLE; - s->flags |=SLAB_TRACE ; - } -return length; -} -SLAB_ATTR (trace); +SLAB_ATTR_RO (trace);static ssize_t red_zone_show (struct kmem_cache *s,char *buf) {@@ -5391,19 +5347,7 @@ static ssize_t failslab_show(struct kmem_cache *s, char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_FAILSLAB )); } - -static ssize_t failslab_store (struct kmem_cache *s,const char *buf, -size_t length) -{ -if (s->refcount >1 ) -return -EINVAL ; - - s->flags &= ~SLAB_FAILSLAB ; -if (buf[0 ] =='1' ) - s->flags |=SLAB_FAILSLAB ; -return length; -} -SLAB_ATTR (failslab); +SLAB_ATTR_RO (failslab);#endif static ssize_t shrink_show (struct kmem_cache *s,char *buf)Signature · 1 line
2.26.2
RFC PATCH 5/5 mm, slub: make reclaim_account attribute read-only
In reply to Vlastimil BabkaThe attribute reflects the SLAB_RECLAIM_ACCOUNT cache flag. It's not clear why this attribute was writable in the first place, as it's tied to how the cache is used by its creator, it's not a user tunable. Furthermore:
- it affects slab merging, but that's not being checked while toggled
- if affects whether __GFP_RECLAIMABLE flag is used to allocate page, but the runtime toggle doesn't update allocflags
- it affects cache_vmstat_idx() so runtime toggle might lead to incosistency of NR_SLAB_RECLAIMABLE and NR_SLAB_UNRECLAIMABLE
Thus make it read-only.
--- mm/slub.c | 11 +---------- 1 file changed, 1 insertion(+), 10 deletions(-)
mm/slub.c +1 -10
@@ -5223,16 +5223,7 @@ static ssize_t reclaim_account_show(struct kmem_cache *s, char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_RECLAIM_ACCOUNT )); } - -static ssize_t reclaim_account_store (struct kmem_cache *s, -const char *buf,size_t length) -{ - s->flags &= ~SLAB_RECLAIM_ACCOUNT ; -if (buf[0 ] =='1' ) - s->flags |=SLAB_RECLAIM_ACCOUNT ; -return length; -} -SLAB_ATTR (reclaim_account); +SLAB_ATTR_RO (reclaim_account);static ssize_t hwcache_align_show (struct kmem_cache *s,char *buf) {
Signature · 1 line
2.26.2
Re: [RFC PATCH 5/5] mm, slub: make reclaim_account attribute read-only
Quoted 10 lines · Vlastimil Babka
The attribute reflects the
SLAB_RECLAIM_ACCOUNTcache flag. It's not clear why this attribute was writable in the first place, as it's tied to how the cache is used by its creator, it's not a user tunable. Furthermore:- it affects slab merging, but that's not being checked while toggled
- if affects whether__GFP_RECLAIMABLEflag is used to allocate page, but the runtime toggle doesn't update allocflags
- it affectscache_vmstat_idx()so runtime toggle might lead to incosistency ofNR_SLAB_RECLAIMABLEandNR_SLAB_UNRECLAIMABLE
Quoted 3 lines
Thus make it read-only.
Signed-off-by Vlastimil Babka <[email protected]>
Signature · 1 line
Kees Cook
Re: [RFC PATCH 5/5] mm, slub: make reclaim_account attribute read-only
In reply to Vlastimil BabkaQuoted 10 lines · Vlastimil Babka
The attribute reflects the
SLAB_RECLAIM_ACCOUNTcache flag. It's not clear why this attribute was writable in the first place, as it's tied to how the cache is used by its creator, it's not a user tunable. Furthermore:- it affects slab merging, but that's not being checked while toggled
- if affects whether__GFP_RECLAIMABLEflag is used to allocate page, but the runtime toggle doesn't update allocflags
- it affectscache_vmstat_idx()so runtime toggle might lead to incosistency ofNR_SLAB_RECLAIMABLEandNR_SLAB_UNRECLAIMABLE
Quoted 3 lines
Thus make it read-only.
Signed-off-by Vlastimil Babka <[email protected]>
Quoted 30 lines
--- mm/slub.c | 11 +---------- 1 file changed, 1 insertion(+), 10 deletions(-)mm/slub.c +1
-10@@ -5223,16 +5223,7 @@ static ssize_t reclaim_account_show(struct kmem_cache *s, char *buf) {return sprintf (buf,"%d\n" , !!(s->flags &SLAB_RECLAIM_ACCOUNT )); } - -static ssize_t reclaim_account_store (struct kmem_cache *s, -const char *buf,size_t length) -{ - s->flags &= ~SLAB_RECLAIM_ACCOUNT ; -if (buf[0 ] =='1' ) - s->flags |=SLAB_RECLAIM_ACCOUNT ; -return length; -} -SLAB_ATTR (reclaim_account); +SLAB_ATTR_RO (reclaim_account);static ssize_t hwcache_align_show (struct kmem_cache *s,char *buf) {Signature · 1 line
2.26.2