Re: linux-next: manual merge of the rcu tree with Linus' tree
On Mon, Jun 18, 2018 at 01:49:47PM +1000, Stephen Rothwell wrote: > Hi Paul, > > On Mon, 18 Jun 2018 13:48:49 +1000 Stephen Rothwell > wrote: > > > > I am still getting this. > > Maybe this was a bit premature :-) Actually, this whole mess was due to me forgetting to do the merge-window reset of rcu/next to the tree actually submitted to mainline. :-/ Please accept my apologies for the resulting hassles! Thanx, Paul
Re: linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul, On Mon, 18 Jun 2018 13:48:49 +1000 Stephen Rothwell wrote: > > I am still getting this. Maybe this was a bit premature :-) -- Cheers, Stephen Rothwell pgpctU4_DnFxM.pgp Description: OpenPGP digital signature
Re: linux-next: manual merge of the rcu tree with Linus' tree
Hi all,
On Tue, 12 Jun 2018 10:46:07 +1000 Stephen Rothwell
wrote:
>
> Today's linux-next merge of the rcu tree got a conflict in:
>
> drivers/iommu/amd_iommu.c
>
> between commit:
>
> 94c793accacd ("iommu/amd: Hide unused iommu_table_lock")
>
> from Linus' tree and commit:
>
> 1df9bb146146 ("EXP iommu: Placeholder for fix in mainline")
>
> from the rcu tree.
I am still getting this.
--
Cheers,
Stephen Rothwell
pgp1L2ooEuUvL.pgp
Description: OpenPGP digital signature
Re: linux-next: manual merge of the rcu tree with Linus' tree
On Tue, Jun 12, 2018 at 10:46:07AM +1000, Stephen Rothwell wrote:
> Hi Paul,
>
> Today's linux-next merge of the rcu tree got a conflict in:
>
> drivers/iommu/amd_iommu.c
>
> between commit:
>
> 94c793accacd ("iommu/amd: Hide unused iommu_table_lock")
>
> from Linus' tree and commit:
>
> 1df9bb146146 ("EXP iommu: Placeholder for fix in mainline")
>
> from the rcu tree.
>
> I fixed it up (they share some code change) and can carry the fix as
> necessary. This is now fixed as far as linux-next is concerned, but any
> non trivial conflicts should be mentioned to your upstream maintainer
> when your tree is submitted for merging. You may also want to consider
> cooperating with the maintainer of the conflicting tree to minimise any
> particularly complex conflicts.
I expect to be dropping my commit as soon as I rebase to v4.18-rc1.
In the meantime, please accept my apologies for the noise!
Thanx, Paul
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in:
drivers/iommu/amd_iommu.c
between commit:
94c793accacd ("iommu/amd: Hide unused iommu_table_lock")
from Linus' tree and commit:
1df9bb146146 ("EXP iommu: Placeholder for fix in mainline")
from the rcu tree.
I fixed it up (they share some code change) and can carry the fix as
necessary. This is now fixed as far as linux-next is concerned, but any
non trivial conflicts should be mentioned to your upstream maintainer
when your tree is submitted for merging. You may also want to consider
cooperating with the maintainer of the conflicting tree to minimise any
particularly complex conflicts.
--
Cheers,
Stephen Rothwell
pgpMiIGk_Cuel.pgp
Description: OpenPGP digital signature
Re: linux-next: manual merge of the rcu tree with Linus' tree
On Mon, May 14, 2018 at 01:46:36PM +1000, Stephen Rothwell wrote:
> Hi all,
>
> Today's linux-next merge of the rcu tree got a conflict in:
>
> drivers/nvme/host/core.c
>
> between commit:
>
> 12d9f07022dc ("nvme: fix use-after-free in nvme_free_ns_head")
>
> from Linus' tree and commit:
>
> d9cf21bae6cf ("nvme: Avoid flush dependency in delete controller flow")
>
> from the rcu tree.
>
> I fixed it up (see below) and can carry the fix as necessary. This
> is now fixed as far as linux-next is concerned, but any non trivial
> conflicts should be mentioned to your upstream maintainer when your tree
> is submitted for merging. You may also want to consider cooperating
> with the maintainer of the conflicting tree to minimise any particularly
> complex conflicts.
Thank you, Stephen!!!
Thanx, Paul
> --
> Cheers,
> Stephen Rothwell
>
> diff --cc drivers/nvme/host/core.c
> index 99b857e5a7a9,c3cea8a29843..
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@@ -351,8 -349,7 +351,8 @@@ static void nvme_free_ns_head(struct kr
> nvme_mpath_remove_disk(head);
> ida_simple_remove(&head->subsys->ns_ida, head->instance);
> list_del_init(&head->entry);
> - cleanup_srcu_struct(&head->srcu);
> + cleanup_srcu_struct_quiesced(&head->srcu);
> +nvme_put_subsystem(head->subsys);
> kfree(head);
> }
>
linux-next: manual merge of the rcu tree with Linus' tree
Hi all,
Today's linux-next merge of the rcu tree got a conflict in:
drivers/nvme/host/core.c
between commit:
12d9f07022dc ("nvme: fix use-after-free in nvme_free_ns_head")
from Linus' tree and commit:
d9cf21bae6cf ("nvme: Avoid flush dependency in delete controller flow")
from the rcu tree.
I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging. You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.
--
Cheers,
Stephen Rothwell
diff --cc drivers/nvme/host/core.c
index 99b857e5a7a9,c3cea8a29843..
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@@ -351,8 -349,7 +351,8 @@@ static void nvme_free_ns_head(struct kr
nvme_mpath_remove_disk(head);
ida_simple_remove(&head->subsys->ns_ida, head->instance);
list_del_init(&head->entry);
- cleanup_srcu_struct(&head->srcu);
+ cleanup_srcu_struct_quiesced(&head->srcu);
+ nvme_put_subsystem(head->subsys);
kfree(head);
}
pgpZj68hxGev0.pgp
Description: OpenPGP digital signature
Re: linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul, On Mon, 25 Sep 2017 20:26:28 -0700 "Paul E. McKenney" wrote: > > This conflict will disappear tomorrow, as I have merged the commit > from Linus's tree in place of mine and have added another commit that > removes the READ_ONCE()s. Same result, but no conflict. ;-) Thanks, I figured something like that would happen. -- Cheers, Stephen Rothwell
Re: linux-next: manual merge of the rcu tree with Linus' tree
On Tue, Sep 26, 2017 at 01:00:18PM +1000, Stephen Rothwell wrote:
> Hi Paul,
>
> Today's linux-next merge of the rcu tree got a conflict in:
>
> kernel/rcu/tree.c
>
> between commit:
>
> 28585a832602 ("rcu: Allow for page faults in NMI handlers")
>
> from Linus' tree and commit:
>
> 3e2baa988b9c ("rcu: Allow for page faults in NMI handlers")
>
> from the rcu tree.
>
> I fixed it up (I just used the rcu tree version) and can carry the fix
> as necessary. This is now fixed as far as linux-next is concerned, but
> any non trivial conflicts should be mentioned to your upstream maintainer
> when your tree is submitted for merging. You may also want to consider
> cooperating with the maintainer of the conflicting tree to minimise any
> particularly complex conflicts.
Hello, Stephen,
This conflict will disappear tomorrow, as I have merged the commit
from Linus's tree in place of mine and have added another commit that
removes the READ_ONCE()s. Same result, but no conflict. ;-)
Thanx, Paul
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in:
kernel/rcu/tree.c
between commit:
28585a832602 ("rcu: Allow for page faults in NMI handlers")
from Linus' tree and commit:
3e2baa988b9c ("rcu: Allow for page faults in NMI handlers")
from the rcu tree.
I fixed it up (I just used the rcu tree version) and can carry the fix
as necessary. This is now fixed as far as linux-next is concerned, but
any non trivial conflicts should be mentioned to your upstream maintainer
when your tree is submitted for merging. You may also want to consider
cooperating with the maintainer of the conflicting tree to minimise any
particularly complex conflicts.
--
Cheers,
Stephen Rothwell
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in:
mm/slab_common.c
between commit:
657dc2f97220 ("slab: remove synchronous rcu_barrier() call in memcg cache
release path")
from Linus' tree and commit:
24b7cb25b8d1 ("mm: Rename SLAB_DESTROY_BY_RCU to SLAB_TYPESAFE_BY_RCU")
from the rcu tree.
I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging. You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.
--
Cheers,
Stephen Rothwell
diff --cc mm/slab_common.c
index 09d0e849b07f,296413c2bbcd..
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@@ -494,55 -458,29 +494,56 @@@ out_unlock
}
EXPORT_SYMBOL(kmem_cache_create);
-static int shutdown_cache(struct kmem_cache *s,
- struct list_head *release, bool *need_rcu_barrier)
+static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
{
- if (__kmem_cache_shutdown(s) != 0)
- return -EBUSY;
+ LIST_HEAD(to_destroy);
+ struct kmem_cache *s, *s2;
+
+ /*
-* On destruction, SLAB_DESTROY_BY_RCU kmem_caches are put on the
++ * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
+ * @slab_caches_to_rcu_destroy list. The slab pages are freed
+ * through RCU and and the associated kmem_cache are dereferenced
+ * while freeing the pages, so the kmem_caches should be freed only
+ * after the pending RCU operations are finished. As rcu_barrier()
+ * is a pretty slow operation, we batch all pending destructions
+ * asynchronously.
+ */
+ mutex_lock(&slab_mutex);
+ list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
+ mutex_unlock(&slab_mutex);
+
+ if (list_empty(&to_destroy))
+ return;
- rcu_barrier();
+ if (s->flags & SLAB_TYPESAFE_BY_RCU)
+ *need_rcu_barrier = true;
- list_move(&s->list, release);
- return 0;
+ list_for_each_entry_safe(s, s2, &to_destroy, list) {
+#ifdef SLAB_SUPPORTS_SYSFS
+ sysfs_slab_release(s);
+#else
+ slab_kmem_cache_release(s);
+#endif
+ }
}
-static void release_caches(struct list_head *release, bool need_rcu_barrier)
+static int shutdown_cache(struct kmem_cache *s)
{
- struct kmem_cache *s, *s2;
+ /* free asan quarantined objects */
+ kasan_cache_shutdown(s);
- if (need_rcu_barrier)
- rcu_barrier();
+ if (__kmem_cache_shutdown(s) != 0)
+ return -EBUSY;
- list_for_each_entry_safe(s, s2, release, list) {
+ memcg_unlink_cache(s);
+ list_del(&s->list);
+
- if (s->flags & SLAB_DESTROY_BY_RCU) {
++ if (s->flags & SLAB_TYPESAFE_BY_RCU) {
+ list_add_tail(&s->list, &slab_caches_to_rcu_destroy);
+ schedule_work(&slab_caches_to_rcu_destroy_work);
+ } else {
#ifdef SLAB_SUPPORTS_SYSFS
- sysfs_slab_remove(s);
+ sysfs_slab_release(s);
#else
slab_kmem_cache_release(s);
#endif
Re: linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
On Wed, 8 Mar 2017 11:46:33 +1100 Stephen Rothwell
wrote:
>
> Today's linux-next merge of the rcu tree got a conflict in:
>
> mm/slab_common.c
>
> between commit:
>
> 657dc2f97220 ("slab: remove synchronous rcu_barrier() call in memcg cache
> release path")
>
> from Linus' tree and commit:
>
> 24b7cb25b8d1 ("mm: Rename SLAB_DESTROY_BY_RCU to SLAB_TYPESAFE_BY_RCU")
>
> from the rcu tree.
>
> I fixed it up (see below) and can carry the fix as necessary. This
> is now fixed as far as linux-next is concerned, but any non trivial
> conflicts should be mentioned to your upstream maintainer when your tree
> is submitted for merging. You may also want to consider cooperating
> with the maintainer of the conflicting tree to minimise any particularly
> complex conflicts.
That resolution was obviously wrong, the correct one is this:
diff --cc mm/slab_common.c
index 09d0e849b07f,296413c2bbcd..
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@@ -494,55 -458,29 +494,55 @@@ out_unlock
}
EXPORT_SYMBOL(kmem_cache_create);
-static int shutdown_cache(struct kmem_cache *s,
- struct list_head *release, bool *need_rcu_barrier)
+static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
{
- if (__kmem_cache_shutdown(s) != 0)
- return -EBUSY;
+ LIST_HEAD(to_destroy);
+ struct kmem_cache *s, *s2;
+
+ /*
-* On destruction, SLAB_DESTROY_BY_RCU kmem_caches are put on the
++ * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
+ * @slab_caches_to_rcu_destroy list. The slab pages are freed
+ * through RCU and and the associated kmem_cache are dereferenced
+ * while freeing the pages, so the kmem_caches should be freed only
+ * after the pending RCU operations are finished. As rcu_barrier()
+ * is a pretty slow operation, we batch all pending destructions
+ * asynchronously.
+ */
+ mutex_lock(&slab_mutex);
+ list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
+ mutex_unlock(&slab_mutex);
+
+ if (list_empty(&to_destroy))
+ return;
- if (s->flags & SLAB_TYPESAFE_BY_RCU)
- *need_rcu_barrier = true;
+ rcu_barrier();
- list_move(&s->list, release);
- return 0;
+ list_for_each_entry_safe(s, s2, &to_destroy, list) {
+#ifdef SLAB_SUPPORTS_SYSFS
+ sysfs_slab_release(s);
+#else
+ slab_kmem_cache_release(s);
+#endif
+ }
}
-static void release_caches(struct list_head *release, bool need_rcu_barrier)
+static int shutdown_cache(struct kmem_cache *s)
{
- struct kmem_cache *s, *s2;
+ /* free asan quarantined objects */
+ kasan_cache_shutdown(s);
- if (need_rcu_barrier)
- rcu_barrier();
+ if (__kmem_cache_shutdown(s) != 0)
+ return -EBUSY;
- list_for_each_entry_safe(s, s2, release, list) {
+ memcg_unlink_cache(s);
+ list_del(&s->list);
+
- if (s->flags & SLAB_DESTROY_BY_RCU) {
++ if (s->flags & SLAB_TYPESAFE_BY_RCU) {
+ list_add_tail(&s->list, &slab_caches_to_rcu_destroy);
+ schedule_work(&slab_caches_to_rcu_destroy_work);
+ } else {
#ifdef SLAB_SUPPORTS_SYSFS
- sysfs_slab_remove(s);
+ sysfs_slab_release(s);
#else
slab_kmem_cache_release(s);
#endif
--
Cheers,
Stephen Rothwell
Re: linux-next: manual merge of the rcu tree with Linus' tree
On Wed, Mar 30, 2016 at 11:32:14AM +1100, Stephen Rothwell wrote:
> Hi Paul,
>
> Today's linux-next merge of the rcu tree got a conflict in:
>
> kernel/rcu/tree.c
>
> between commit:
>
> abedf8e2419f ("rcu: Use simple wait queues where possible in rcutree")
>
> from Linus' tree and commit:
>
> 08cace5914ea ("DIAGS: Crude exploratory hack")
>
> from the rcu tree.
>
> I fixed it up (I used the rcu tree version) and can carry the fix as
> necessary. This is now fixed as far as linux-next is concerned, but any
> non trivial conflicts should be mentioned to your upstream maintainer
> when your tree is submitted for merging. You may also want to consider
> cooperating with the maintainer of the conflicting tree to minimise any
> particularly complex conflicts.
>
> I really don't think that that rcu tree patch should be in linux-next,
> right?
Right you are! I will rework to get the diagnostics out of your way.
-Might- have helped find one of the bugs, but don't look now...
Thanx, Paul
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in:
kernel/rcu/tree.c
between commit:
abedf8e2419f ("rcu: Use simple wait queues where possible in rcutree")
from Linus' tree and commit:
08cace5914ea ("DIAGS: Crude exploratory hack")
from the rcu tree.
I fixed it up (I used the rcu tree version) and can carry the fix as
necessary. This is now fixed as far as linux-next is concerned, but any
non trivial conflicts should be mentioned to your upstream maintainer
when your tree is submitted for merging. You may also want to consider
cooperating with the maintainer of the conflicting tree to minimise any
particularly complex conflicts.
I really don't think that that rcu tree patch should be in linux-next,
right?
--
Cheers,
Stephen Rothwell
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in:
kernel/rcu/rcutorture.c
between commit:
25528213fe9f ("tags: Fix DEFINE_PER_CPU expansions")
from Linus' tree and commit:
0b36a04e0461 ("rcutorture: Remove redundant initialization to zero")
from the rcu tree.
I fixed it up (the latter also did what the former did) and can carry the
fix as necessary. This is now fixed as far as linux-next is concerned,
but any non trivial conflicts should be mentioned to your upstream
maintainer when your tree is submitted for merging. You may also want
to consider cooperating with the maintainer of the conflicting tree to
minimise any particularly complex conflicts.
--
Cheers,
Stephen Rothwell
Re: linux-next: manual merge of the rcu tree with Linus' tree
On Thu, Nov 29, 2012 at 02:06:51PM +1100, Stephen Rothwell wrote:
> Hi Paul,
>
> Today's linux-next merge of the rcu tree got a conflict in
> arch/x86/kernel/ptrace.c between commit cb57a2b4cff7 ("x86-32: Export
> kernel_stack_pointer() for modules") from Linus' tree and commit
> 98dbec158343 ("context_tracking: New context tracking susbsystem") from
> the rcu tree.
>
> I fixed it up (see below) and can carry the fix as necessary (no action
> is required).
Looks good, thank you!
Thanx, Paul
> --
> Cheers,
> Stephen Rothwells...@canb.auug.org.au
>
> diff --cc arch/x86/kernel/ptrace.c
> index 974b67e,65b88a5..000
> --- a/arch/x86/kernel/ptrace.c
> +++ b/arch/x86/kernel/ptrace.c
> @@@ -21,8 -21,7 +21,8 @@@
> #include
> #include
> #include
> - #include
> +#include
> + #include
>
> #include
> #include
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in
arch/x86/kernel/ptrace.c between commit cb57a2b4cff7 ("x86-32: Export
kernel_stack_pointer() for modules") from Linus' tree and commit
98dbec158343 ("context_tracking: New context tracking susbsystem") from
the rcu tree.
I fixed it up (see below) and can carry the fix as necessary (no action
is required).
--
Cheers,
Stephen Rothwells...@canb.auug.org.au
diff --cc arch/x86/kernel/ptrace.c
index 974b67e,65b88a5..000
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@@ -21,8 -21,7 +21,8 @@@
#include
#include
#include
- #include
+#include
+ #include
#include
#include
pgpgYanL0fZ9b.pgp
Description: PGP signature
linux-next: manual merge of the rcu tree with Linus' tree
Hi Paul,
Today's linux-next merge of the rcu tree got a conflict in
include/linux/sched.h between commit 907aed48f65e ("mm: allow PF_MEMALLOC
from softirq context") from Linus' tree and commit 46fc4e7c01b7 ("rcu:
Switch task's syscall hooks on context switch") from the rcu tree.
Just context changes. I fixed it up (see below) and can carry the fix as
necessary.
--
Cheers,
Stephen Rothwells...@canb.auug.org.au
diff --cc include/linux/sched.h
index b8c8664,a094959..000
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@@ -1886,13 -1879,14 +1886,21 @@@ static inline void rcu_copy_process(str
#endif
+static inline void tsk_restore_flags(struct task_struct *task,
+ unsigned long orig_flags, unsigned long flags)
+{
+ task->flags &= ~flags;
+ task->flags |= orig_flags & flags;
+}
+
+ static inline void rcu_switch(struct task_struct *prev,
+ struct task_struct *next)
+ {
+ #ifdef CONFIG_RCU_USER_QS
+ rcu_user_hooks_switch(prev, next);
+ #endif
+ }
+
#ifdef CONFIG_SMP
extern void do_set_cpus_allowed(struct task_struct *p,
const struct cpumask *new_mask);
pgphxu3tUAmBM.pgp
Description: PGP signature
