Consider one situation:
The app have two vmas which mbind() to node 1 and node3 respectively,
and its cpuset.mems is 0-3, now set its cpuset.mems to 1,3, according
to current bitmap_remap(), we got:
1 => 3
3 => 3
This maybe confused because node 1,3 have already in the new settiing
region but both nodes are binded to the same node 3 now.
Actually we found the situation on a very old libvirt and qemu, but
this can be easily reproduced in the current kernel, so we try to fix
it.
A possible fix way is to ignore the bits in @src have already existed
in @new.
Signed-off-by: Yi Wang <[email protected]>
---
lib/bitmap.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/bitmap.c b/lib/bitmap.c
index b18e31ea6e66..b77bf1b3852e 100644
--- a/lib/bitmap.c
+++ b/lib/bitmap.c
@@ -1006,8 +1006,8 @@ unsigned int bitmap_ord_to_pos(const unsigned long *buf, unsigned int ord, unsig
* @dst point to the same location, then this routine copies @src
* to @dst.
*
- * The positions of unset bits in @old are mapped to themselves
- * (the identify map).
+ * The positions of unset bits in @old or bits in @src have already
+ * existed in @new are mapped to themselves (the identify map).
*
* Apply the above specified mapping to @src, placing the result in
* @dst, clearing any bits previously set in @dst.
@@ -1033,7 +1033,7 @@ void bitmap_remap(unsigned long *dst, const unsigned long *src,
for_each_set_bit(oldbit, src, nbits) {
int n = bitmap_pos_to_ord(old, oldbit, nbits);
- if (n < 0 || w == 0)
+ if (n < 0 || w == 0 || test_bit(oldbit, new))
set_bit(oldbit, dst); /* identity map */
else
set_bit(bitmap_ord_to_pos(new, n % w, nbits), dst);
--
2.33.0.rc0.dirty
On Mon, Jun 13, 2022 at 4:31 AM Yi Wang <[email protected]> wrote:
>
> Consider one situation:
>
> The app have two vmas which mbind() to node 1 and node3 respectively,
> and its cpuset.mems is 0-3, now set its cpuset.mems to 1,3, according
> to current bitmap_remap(), we got:
>
> 1 => 3
> 3 => 3
>
> This maybe confused because node 1,3 have already in the new settiing
> region but both nodes are binded to the same node 3 now.
>
> Actually we found the situation on a very old libvirt and qemu, but
> this can be easily reproduced in the current kernel, so we try to fix
> it.
>
> A possible fix way is to ignore the bits in @src have already existed
> in @new.
>
> Signed-off-by: Yi Wang <[email protected]>
> ---
> lib/bitmap.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/lib/bitmap.c b/lib/bitmap.c
> index b18e31ea6e66..b77bf1b3852e 100644
> --- a/lib/bitmap.c
> +++ b/lib/bitmap.c
> @@ -1006,8 +1006,8 @@ unsigned int bitmap_ord_to_pos(const unsigned long *buf, unsigned int ord, unsig
> * @dst point to the same location, then this routine copies @src
> * to @dst.
> *
> - * The positions of unset bits in @old are mapped to themselves
> - * (the identify map).
> + * The positions of unset bits in @old or bits in @src have already
> + * existed in @new are mapped to themselves (the identify map).
> *
> * Apply the above specified mapping to @src, placing the result in
> * @dst, clearing any bits previously set in @dst.
> @@ -1033,7 +1033,7 @@ void bitmap_remap(unsigned long *dst, const unsigned long *src,
> for_each_set_bit(oldbit, src, nbits) {
> int n = bitmap_pos_to_ord(old, oldbit, nbits);
>
> - if (n < 0 || w == 0)
> + if (n < 0 || w == 0 || test_bit(oldbit, new))
> set_bit(oldbit, dst); /* identity map */
> else
> set_bit(bitmap_ord_to_pos(new, n % w, nbits), dst);
> --
> 2.33.0.rc0.dirty
Regarding the original problem - can you please confirm that
it's reproduced on current kernels, show the execution path etc.
From what I see on modern kernel, the only user of nodes_remap()
is mpol_rebind_nodemask(). Is that the correct path?
Anyways, as per name, bitmap_remap() is intended to change bit
positions, and it doesn't look wrong if it does so.
This is not how the function is supposed to work. For example,
old: 00111000
new: 00011100
means:
old: 00111 000
|| \\\|||
new: 000 11100
And after this patch it would be:
old: 001 11000
|| \|||||
new: 000 11100
Which is not the same, right?
If mpol_rebind() wants to keep previous relations, then according to
the comment:
* The positions of unset bits in @old are mapped to themselves
* (the identify map).
, you can just clear @old bits that already have good relations
you'd like to preserve.
Thanks,
Yury
Hi Yury,
Thanks for your quick and clear response!
> On Mon, Jun 13, 2022 at 4:31 AM Yi Wang <[email protected]> wrote:
> >
> > Consider one situation:
> >
> > The app have two vmas which mbind() to node 1 and node3 respectively,
> > and its cpuset.mems is 0-3, now set its cpuset.mems to 1,3, according
> > to current bitmap_remap(), we got:
>
> Regarding the original problem - can you please confirm that
> it's reproduced on current kernels, show the execution path etc.
> From what I see on modern kernel, the only user of nodes_remap()
> is mpol_rebind_nodemask(). Is that the correct path?
Yes, it's mpol_rebind_nodemask() calls nodes_remap() from
mpol_rebind_policy(). The stacks are as follow:
[ 290.836747] bitmap_remap+0x84/0xe0
[ 290.836753] mpol_rebind_nodemask+0x64/0x2a0
[ 290.836764] mpol_rebind_mm+0x3a/0x90
[ 290.836768] update_tasks_nodemask+0x8a/0x1e0
[ 290.836774] cpuset_write_resmask+0x563/0xa00
[ 290.836780] cgroup_file_write+0x81/0x150
[ 290.836784] kernfs_fop_write_iter+0x12d/0x1c0
[ 290.836791] new_sync_write+0x109/0x190
[ 290.836800] vfs_write+0x218/0x2a0
[ 290.836809] ksys_write+0x59/0xd0
[ 290.836812] do_syscall_64+0x37/0x80
[ 290.836818] entry_SYSCALL_64_after_hwframe+0x46/0xb0
To reproduce this situation, I write a program which seems like this:
unsigned int flags = MAP_PRIVATE | MAP_ANONYMOUS;
unsigned long size = 262144 << 12;
unsigned long node1 = 2; // node 1
unsigned long node2 = 8; // node 3
p1 = vma1 = mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
p2 = vma2 = mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
assert(!mbind(vma1, size, MPOL_BIND, &node1, MAX_NODES, MPOL_MF_STRICT | MPOL_MF_MOVE));
assert(!mbind(vma2, size, MPOL_BIND, &node2, MAX_NODES, MPOL_MF_STRICT | MPOL_MF_MOVE));
Start the program whos name is mbind_tester, and do follow steps:
mkdir && cd /sys/fs/cgroup/cpuset/mbind
echo 0-31 > cpuset.cpus
echo 0-3 > cpuset.mems
cat /proc/`pidof mbind_tester`/numa_maps |grep bind -w
7ff73e200000 bind:3 anon=262144 dirty=262144 active=0 N3=262144 kernelpagesize_kB=4
7ff77e200000 bind:1 anon=262144 dirty=262144 active=0 N1=262144 kernelpagesize_kB=4
echo 1,3 > cpuset.mems
cat /proc/`pidof mbind_tester`/numa_maps |grep bind -w
7ff73e200000 bind:3 anon=262144 dirty=262144 active=0 N3=262144 kernelpagesize_kB=4
7ff77e200000 bind:3 anon=262144 dirty=262144 active=0 N1=262144 kernelpagesize_kB=4
As you see, after set cpuset.mems to 1,3, the nodes which one of vma
binded to changed from 1 to 3.
This maybe confused, the original nodes binded is 1, after modify
cpuset.mems to 1,3 which include the node 3, it changed to 3...
>
> Anyways, as per name, bitmap_remap() is intended to change bit
> positions, and it doesn't look wrong if it does so.
>
> This is not how the function is supposed to work. For example,
> old: 00111000
> new: 00011100
>
> means:
> old: 00111 000
> || \\\|||
> new: 000 11100
>
> And after this patch it would be:
> old: 001 11000
> || \|||||
> new: 000 11100
>
> Which is not the same, right?
Right.
Actually this is what makes me embarrassed. If we want to fix this
situtation, we can:
- change the bitmap_remap() as this patch did, but this changed the
behavior of this routine which looks does the right thing. One good
news is this function is only called by mpol_rebind_nodemask().
- don't change the bitmap_remap(), to be honest, I didn't figure out
a way. Any suggestions?
>
> If mpol_rebind() wants to keep previous relations, then according to
> the comment:
> * The positions of unset bits in @old are mapped to themselves
> * (the identify map).
>
> , you can just clear @old bits that already have good relations
> you'd like to preserve.
Actually this does not work for me :)
In the example above, if set cpuset.mems to 0,2 firstly, the nodes
binds will change from 1 to 2. And then set cpuset.mems to 1,3, it will
change from 2 to 3 again.
---
Best wishes
Yi Wang
+ Andrew Morton <[email protected]>
+ [email protected]
On Mon, Jun 13, 2022 at 8:45 PM <[email protected]> wrote:
>
> Hi Yury,
>
> Thanks for your quick and clear response!
>
> > On Mon, Jun 13, 2022 at 4:31 AM Yi Wang <[email protected]> wrote:
> > >
> > > Consider one situation:
> > >
> > > The app have two vmas which mbind() to node 1 and node3 respectively,
> > > and its cpuset.mems is 0-3, now set its cpuset.mems to 1,3, according
> > > to current bitmap_remap(), we got:
> >
> > Regarding the original problem - can you please confirm that
> > it's reproduced on current kernels, show the execution path etc.
> > From what I see on modern kernel, the only user of nodes_remap()
> > is mpol_rebind_nodemask(). Is that the correct path?
>
> Yes, it's mpol_rebind_nodemask() calls nodes_remap() from
> mpol_rebind_policy(). The stacks are as follow:
> [ 290.836747] bitmap_remap+0x84/0xe0
> [ 290.836753] mpol_rebind_nodemask+0x64/0x2a0
> [ 290.836764] mpol_rebind_mm+0x3a/0x90
> [ 290.836768] update_tasks_nodemask+0x8a/0x1e0
> [ 290.836774] cpuset_write_resmask+0x563/0xa00
> [ 290.836780] cgroup_file_write+0x81/0x150
> [ 290.836784] kernfs_fop_write_iter+0x12d/0x1c0
> [ 290.836791] new_sync_write+0x109/0x190
> [ 290.836800] vfs_write+0x218/0x2a0
> [ 290.836809] ksys_write+0x59/0xd0
> [ 290.836812] do_syscall_64+0x37/0x80
> [ 290.836818] entry_SYSCALL_64_after_hwframe+0x46/0xb0
>
> To reproduce this situation, I write a program which seems like this:
> unsigned int flags = MAP_PRIVATE | MAP_ANONYMOUS;
> unsigned long size = 262144 << 12;
> unsigned long node1 = 2; // node 1
> unsigned long node2 = 8; // node 3
>
> p1 = vma1 = mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
> p2 = vma2 = mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
>
> assert(!mbind(vma1, size, MPOL_BIND, &node1, MAX_NODES, MPOL_MF_STRICT | MPOL_MF_MOVE));
> assert(!mbind(vma2, size, MPOL_BIND, &node2, MAX_NODES, MPOL_MF_STRICT | MPOL_MF_MOVE));
>
> Start the program whos name is mbind_tester, and do follow steps:
>
> mkdir && cd /sys/fs/cgroup/cpuset/mbind
> echo 0-31 > cpuset.cpus
> echo 0-3 > cpuset.mems
>
> cat /proc/`pidof mbind_tester`/numa_maps |grep bind -w
> 7ff73e200000 bind:3 anon=262144 dirty=262144 active=0 N3=262144 kernelpagesize_kB=4
> 7ff77e200000 bind:1 anon=262144 dirty=262144 active=0 N1=262144 kernelpagesize_kB=4
>
> echo 1,3 > cpuset.mems
> cat /proc/`pidof mbind_tester`/numa_maps |grep bind -w
> 7ff73e200000 bind:3 anon=262144 dirty=262144 active=0 N3=262144 kernelpagesize_kB=4
> 7ff77e200000 bind:3 anon=262144 dirty=262144 active=0 N1=262144 kernelpagesize_kB=4
>
> As you see, after set cpuset.mems to 1,3, the nodes which one of vma
> binded to changed from 1 to 3.
>
> This maybe confused, the original nodes binded is 1, after modify
> cpuset.mems to 1,3 which include the node 3, it changed to 3...
Ok, thanks for the reproducer. I'll take a look at it closer to the weekend.
> > Anyways, as per name, bitmap_remap() is intended to change bit
> > positions, and it doesn't look wrong if it does so.
> >
> > This is not how the function is supposed to work. For example,
> > old: 00111000
> > new: 00011100
> >
> > means:
> > old: 00111 000
> > || \\\|||
> > new: 000 11100
> >
> > And after this patch it would be:
> > old: 001 11000
> > || \|||||
> > new: 000 11100
> >
> > Which is not the same, right?
>
> Right.
So, we both agree that bitmap_remap() works as advertised. This is
good. Let's try figuring out a solution without touching it.
> Actually this is what makes me embarrassed. If we want to fix this
> situtation, we can:
>
> - change the bitmap_remap() as this patch did, but this changed the
> behavior of this routine which looks does the right thing. One good
> news is this function is only called by mpol_rebind_nodemask().
There are users of bitmap_remap() in drivers/gpio/gpio-xilinx.c
> - don't change the bitmap_remap(), to be honest, I didn't figure out
> a way. Any suggestions?
I haven't had a chance to play with it (because of my dayjob), but I
have a strong feeling that the proper solution should come from
existing functionality.
Did you experiment with MPOL_F_{STATIC,RELATIVE}_NODES?
Those flags enable nodes_and() and mpol_relative_nodemask()
paths correspondingly.
> > If mpol_rebind() wants to keep previous relations, then according to
> > the comment:
> > * The positions of unset bits in @old are mapped to themselves
> > * (the identify map).
> >
> > , you can just clear @old bits that already have good relations
> > you'd like to preserve.
>
> Actually this does not work for me :)
What I suggested is:
328 node_clear(1, pol->w.cpuset_mems_allowed);
329 node_clear(3, pol->w.cpuset_mems_allowed);
330 nodes_remap(tmp, pol->nodes, pol->w.cpuset_mems_allowed,
331 *nodes);
332 pol->w.cpuset_mems_allowed = *nodes;
> In the example above, if set cpuset.mems to 0,2 firstly, the nodes
> binds will change from 1 to 2. And then set cpuset.mems to 1,3, it will
> change from 2 to 3 again.
I bet you can find a sequence that will finally give you the desired binding.
And probably somebody does this black magic in production. For me it
looks just scary. Can you try those static/relative flags, and if it doesn't
work, we'll have to invent another policy for nodes binding.
Adding Andrew and linux-mm, as it's definitely beyond bitmaps scope.
Thanks,
Yury
> ---
> Best wishes
> Yi Wang
On Tue, Jun 14, 2022 at 09:40:08AM -0700, Yury Norov wrote:
> On Mon, Jun 13, 2022 at 8:45 PM <[email protected]> wrote:
> > > On Mon, Jun 13, 2022 at 4:31 AM Yi Wang <[email protected]> wrote:
...
> > > Anyways, as per name, bitmap_remap() is intended to change bit
> > > positions, and it doesn't look wrong if it does so.
Haven't read the discussion in full, but saw the function here and my 2 cents
are the following:
1) if we don't have tests for bitmap_remap() / bitmap_bitremap(), we should;
2) these functions are used at least in one GPIO driver, so it would be nice
to not touch them if they are not broken (see 1 above).
--
With Best Regards,
Andy Shevchenko
> On Mon, Jun 13, 2022 at 8:45 PM <[email protected]> wrote:
> >
> >
> > - change the bitmap_remap() as this patch did, but this changed the
> > behavior of this routine which looks does the right thing. One good
> > news is this function is only called by mpol_rebind_nodemask().
>
> There are users of bitmap_remap() in drivers/gpio/gpio-xilinx.c
>
> > - don't change the bitmap_remap(), to be honest, I didn't figure out
> > a way. Any suggestions?
>
> I haven't had a chance to play with it (because of my dayjob), but I
> have a strong feeling that the proper solution should come from
> existing functionality.
>
> Did you experiment with MPOL_F_{STATIC,RELATIVE}_NODES?
> Those flags enable nodes_and() and mpol_relative_nodemask()
> paths correspondingly.
>
You are right, adding MPOL_F_STATIC_NODES can fix this situation. So it
seems that we don't need to do anything in kernel now.
Many thanks for your time :)
> What I suggested is:
>
> 328 node_clear(1, pol->w.cpuset_mems_allowed);
> 329 node_clear(3, pol->w.cpuset_mems_allowed);
> 330 nodes_remap(tmp, pol->nodes, pol->w.cpuset_mems_allowed,
> 331 *nodes);
> 332 pol->w.cpuset_mems_allowed = *nodes;
>
> > In the example above, if set cpuset.mems to 0,2 firstly, the nodes
> > binds will change from 1 to 2. And then set cpuset.mems to 1,3, it will
> > change from 2 to 3 again.
---
Best wishes
Yi Wang