Attachment 1973843 Details for Bug 2213267 – Patch - fix expire retry looping

[patch] Patch - fix expire retry looping

autofs-5.1.8-fix-expire-retry-looping.patch (text/plain), 3.15 KB, created by Ian Kent on 2023-07-03 08:25:45 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Ian Kent

Created: 2023-07-03 08:25:45 UTC

Size: 3.15 KB

patch

obsolete

>autofs-5.1.8 - fix expire retry looping
>
>From: Ian Kent <raven@themaw.net>
>
>Commit aa6da48d1 (autofs-5.1.7 - eliminate count_mounts() from
>expire_proc_indirect()) stopped using the count_mounts() function
>in indirect mount expires because it can be a significant overhead
>and shouldn't be needed if the kernel expire dentry selection works
>as it should.
>
>Unfortunately there is a case where it doesn't work properly, when
>a USR1 signal is sent to the automount process it is meant to expire
>mounts regardless of the expire timeout. In this case if a mount has
>been propagated to a mount namespace and is held busy the mount will
>fail to umount and because setting the last used field of the mount
>dentry doesn't prevent the mount dentry from being selected for expire
>again immediately in this case automount will look continually.
>
>The problem occurs because the the kernel doesn't know how to check
>these propagated mounts for busyness and the init namespace automount
>process tries to expire the mount but fails and continues trying to
>expire the mount because the expire function assumes only mounts that
>are not busy will be selected for expire.
>
>Signed-off-by: Ian Kent <raven@themaw.net>
>---
> CHANGELOG           |    1 +
> daemon/indirect.c   |   13 ++++++++++++-
> include/automount.h |    2 +-
> 3 files changed, 14 insertions(+), 2 deletions(-)
>
>diff --git a/CHANGELOG b/CHANGELOG
>index a2b59eac..00c3278a 100644
>--- a/CHANGELOG
>+++ b/CHANGELOG
>@@ -92,6 +92,7 @@
> - make open files limit configurable.
> - fix some sss error return cases.
> - fix incorrect matching of cached wildcard key.
>+- fix expire retry looping.
> 
> 19/10/2021 autofs-5.1.8
> - add xdr_exports().
>diff --git a/daemon/indirect.c b/daemon/indirect.c
>index 204fa076..0f7b620b 100644
>--- a/daemon/indirect.c
>+++ b/daemon/indirect.c
>@@ -343,6 +343,7 @@ void *expire_proc_indirect(void *arg)
> 	int offsets, submnts, count;
> 	int ioctlfd, cur_state;
> 	int status, ret, left;
>+	int retries;
> 
> 	ea = (struct expire_args *) arg;
> 
>@@ -490,9 +491,19 @@ void *expire_proc_indirect(void *arg)
> 	 * If there are no more real mounts left we could still
> 	 * have some offset mounts with no '/' offset or symlinks
> 	 * so we need to umount or unlink them here.
>+	 *
>+	 * The dentry info last_used field is set to 'now' when a
>+	 * dentry is selected for expire so that it isn't immediately
>+	 * selected again if the expire fails. But this can't work
>+	 * for immediate expires so the count_mounts() function must
>+	 * be used to limit the number of expire iterations.
> 	 */
> 	pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &cur_state);
>-	while (1) {
>+	if (how == AUTOFS_EXP_IMMEDIATE)
>+		retries = count_mounts(ap, ap->path, ap->dev);
>+	else
>+		retries = -1;
>+	while (retries--) {
> 		ret = ops->expire(ap->logopt, ap->ioctlfd, ap->path, how);
> 		if (ret != 0 && errno == EAGAIN)
> 			break;
>diff --git a/include/automount.h b/include/automount.h
>index 77ff0c70..9548db83 100644
>--- a/include/automount.h
>+++ b/include/automount.h
>@@ -146,7 +146,7 @@ struct autofs_point;
> #define NEGATIVE_TIMEOUT	10
> #define POSITIVE_TIMEOUT	120
> #define UMOUNT_RETRIES		16
>-#define EXPIRE_RETRIES		3
>+#define EXPIRE_RETRIES		1
> 
> struct mapent_cache {
> 	pthread_rwlock_t rwlock;

Actions: View | Diff

Attachments on bug 2213267: 1973843