DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] [PATCH] mem: fix free segment when using huge-unlink option
@ 2021-02-22 10:41 Roy Shterman
  2021-02-22 15:53 ` Burakov, Anatoly
  0 siblings, 1 reply; 3+ messages in thread
From: Roy Shterman @ 2021-02-22 10:41 UTC (permalink / raw)
  To: anatoly.burakov; +Cc: dev, yuval, aviv.bendavid, Roy Shterman

When using huge_unlink we unlink the segment right
after allocation. Although we unlink the file we keep
the fd in fd_list so file still exist just the path deleted.
When freeing the hugepage we need to close the fd and assign
it with (-1) in fd_list for the page to be released.

The current flow fails rte_malloc in the following flow when working
with --huge-unlink option:
1. alloc_seg() for segment A -
    We allocate segment, unlink the path to the segment
    and keep the file descriptor in fd_list.
2. free_seg() for segment A -
    We clear the segment metadata and return - without closing fd
    or assigning (-1) in fd list.
3. alloc_seg() for segment A again -
    We find segment A as available, try to allocate it,
    find the old fd in fd_list try to unlink it
    as part of alloc_seg() but failed because path doesn't exist.

The impact of such error is falsly failing rte_malloc()
although we have hugepages available.

Fixes: d435aad37da7 ("mem: support --huge-unlink mode")

Signed-off-by: Roy Shterman <roy.shterman@vastdata.com>
---
 lib/librte_eal/linux/eal_memalloc.c | 14 ++------------
 1 file changed, 2 insertions(+), 12 deletions(-)

diff --git a/lib/librte_eal/linux/eal_memalloc.c b/lib/librte_eal/linux/eal_memalloc.c
index 6dc1b2bae..c590d6043 100644
--- a/lib/librte_eal/linux/eal_memalloc.c
+++ b/lib/librte_eal/linux/eal_memalloc.c
@@ -709,7 +709,6 @@ free_seg(struct rte_memseg *ms, struct hugepage_info *hi,
 	uint64_t map_offset;
 	char path[PATH_MAX];
 	int fd, ret = 0;
-	bool exit_early;
 	const struct internal_config *internal_conf =
 		eal_get_internal_configuration();
 
@@ -725,17 +724,8 @@ free_seg(struct rte_memseg *ms, struct hugepage_info *hi,
 
 	eal_mem_set_dump(ms->addr, ms->len, false);
 
-	exit_early = false;
-
 	/* if we're using anonymous hugepages, nothing to be done */
-	if (internal_conf->in_memory && !memfd_create_supported)
-		exit_early = true;
-
-	/* if we've already unlinked the page, nothing needs to be done */
-	if (!internal_conf->in_memory && internal_conf->hugepage_unlink)
-		exit_early = true;
-
-	if (exit_early) {
+	if (internal_conf->in_memory && !memfd_create_supported) {
 		memset(ms, 0, sizeof(*ms));
 		return 0;
 	}
@@ -761,7 +751,7 @@ free_seg(struct rte_memseg *ms, struct hugepage_info *hi,
 		/* if we're able to take out a write lock, we're the last one
 		 * holding onto this page.
 		 */
-		if (!internal_conf->in_memory) {
+		if (!internal_conf->in_memory && !internal_conf->hugepage_unlink) {
 			ret = lock(fd, LOCK_EX);
 			if (ret >= 0) {
 				/* no one else is using this page */
-- 
2.24.1


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [dpdk-dev] [PATCH] mem: fix free segment when using huge-unlink option
  2021-02-22 10:41 [dpdk-dev] [PATCH] mem: fix free segment when using huge-unlink option Roy Shterman
@ 2021-02-22 15:53 ` Burakov, Anatoly
  2021-02-28 13:21   ` Roy Shterman
  0 siblings, 1 reply; 3+ messages in thread
From: Burakov, Anatoly @ 2021-02-22 15:53 UTC (permalink / raw)
  To: Roy Shterman; +Cc: dev, yuval, aviv.bendavid

On 22-Feb-21 10:41 AM, Roy Shterman wrote:
> When using huge_unlink we unlink the segment right
> after allocation. Although we unlink the file we keep
> the fd in fd_list so file still exist just the path deleted.
> When freeing the hugepage we need to close the fd and assign
> it with (-1) in fd_list for the page to be released.
> 
> The current flow fails rte_malloc in the following flow when working
> with --huge-unlink option:
> 1. alloc_seg() for segment A -
>      We allocate segment, unlink the path to the segment
>      and keep the file descriptor in fd_list.
> 2. free_seg() for segment A -
>      We clear the segment metadata and return - without closing fd
>      or assigning (-1) in fd list.
> 3. alloc_seg() for segment A again -
>      We find segment A as available, try to allocate it,
>      find the old fd in fd_list try to unlink it
>      as part of alloc_seg() but failed because path doesn't exist.
> 
> The impact of such error is falsly failing rte_malloc()
> although we have hugepages available.
> 
> Fixes: d435aad37da7 ("mem: support --huge-unlink mode")
> 
> Signed-off-by: Roy Shterman <roy.shterman@vastdata.com>

Cc: stable@dpdk.org

Provisionally, patch looks fine, but i'll have to have a closer look.

-- 
Thanks,
Anatoly

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [dpdk-dev] [PATCH] mem: fix free segment when using huge-unlink option
  2021-02-22 15:53 ` Burakov, Anatoly
@ 2021-02-28 13:21   ` Roy Shterman
  0 siblings, 0 replies; 3+ messages in thread
From: Roy Shterman @ 2021-02-28 13:21 UTC (permalink / raw)
  To: Burakov, Anatoly; +Cc: dev, Yuval Mintz, Aviv Ben David

On Mon, Feb 22, 2021 at 5:53 PM Burakov, Anatoly <anatoly.burakov@intel.com>
wrote:

> On 22-Feb-21 10:41 AM, Roy Shterman wrote:
> > When using huge_unlink we unlink the segment right
> > after allocation. Although we unlink the file we keep
> > the fd in fd_list so file still exist just the path deleted.
> > When freeing the hugepage we need to close the fd and assign
> > it with (-1) in fd_list for the page to be released.
> >
> > The current flow fails rte_malloc in the following flow when working
> > with --huge-unlink option:
> > 1. alloc_seg() for segment A -
> >      We allocate segment, unlink the path to the segment
> >      and keep the file descriptor in fd_list.
> > 2. free_seg() for segment A -
> >      We clear the segment metadata and return - without closing fd
> >      or assigning (-1) in fd list.
> > 3. alloc_seg() for segment A again -
> >      We find segment A as available, try to allocate it,
> >      find the old fd in fd_list try to unlink it
> >      as part of alloc_seg() but failed because path doesn't exist.
> >
> > The impact of such error is falsly failing rte_malloc()
> > although we have hugepages available.
> >
> > Fixes: d435aad37da7 ("mem: support --huge-unlink mode")
> >
> > Signed-off-by: Roy Shterman <roy.shterman@vastdata.com>
>
> Cc: stable@dpdk.org
>
> Provisionally, patch looks fine, but i'll have to have a closer look.
>

Hi Anatoly,

Do I need to send this patch also to stable or it will happen automatically
if the patch will reach the next release candidate?
Also I wonder if you had more time to review this one?


>
> --
> Thanks,
> Anatoly
>

Thanks,
Roy

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2021-02-28 14:38 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-02-22 10:41 [dpdk-dev] [PATCH] mem: fix free segment when using huge-unlink option Roy Shterman
2021-02-22 15:53 ` Burakov, Anatoly
2021-02-28 13:21   ` Roy Shterman

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror http://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ http://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git