[gentoo-commits] gentoo-x86 commit in sys-fs/zfs-kmod/files: zfs-kmod-0.6.0_rc9-remove-pfmalloc-3-of-3.patch zfs-kmod-0.6.0_rc9-remove-pfmalloc-1-of-3.patch zfs-kmod-0.6.0_rc9-remove-pfmalloc-2-of-3.patch

[Richard Yao (ryao)]

ryao        12/08/22 07:28:13

  Added:                zfs-kmod-0.6.0_rc9-remove-pfmalloc-3-of-3.patch
                        zfs-kmod-0.6.0_rc9-remove-pfmalloc-1-of-3.patch
                        zfs-kmod-0.6.0_rc9-remove-pfmalloc-2-of-3.patch
  Log:
  Ebuild to provide kernel module for sys-fs/zfs
  
  (Portage version: 2.1.11.9/cvs/Linux x86_64)

Revision  Changes    Path
1.1                  sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-3-of-3.patch

file : http://sources.gentoo.org/viewvc.cgi/gentoo-x86/sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-3-of-3.patch?rev=1.1&view=markup
plain: http://sources.gentoo.org/viewvc.cgi/gentoo-x86/sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-3-of-3.patch?rev=1.1&content-type=text/plain

Index: zfs-kmod-0.6.0_rc9-remove-pfmalloc-3-of-3.patch
===================================================================
From a1c889dd6fb04d70439074a9399c9ea8f29f2cdb Mon Sep 17 00:00:00 2001
From: Richard Yao <ryao@×××××××××××××.edu>
Date: Wed, 16 May 2012 18:16:02 -0400
Subject: [PATCH] Revert Disable direct reclaim for z_wr_* threads

This commit used PF_MEMALLOC to prevent a memory reclaim deadlock.
However, commit 49be0ccf1fdc2ce852271d4d2f8b7a9c2c4be6db eliminated
the invocation of __cv_init(), which was the cause of the deadlock.
PF_MEMALLOC has the side effect of permitting pages from ZONE_DMA
to be allocated.  The use of PF_MEMALLOC was found to cause stability
problems when doing swap on zvols. Since this technique is known to
cause problems and no longer fixes anything, we revert it.

Signed-off-by: Richard Yao <ryao@×××××××××××××.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@××××.gov>
Issue #726
---
 include/sys/zfs_context.h |    1 -
 module/zfs/spa.c          |    9 +++------
 2 files changed, 3 insertions(+), 7 deletions(-)

diff --git a/include/sys/zfs_context.h b/include/sys/zfs_context.h
index e4af6fc..77dded3 100644
--- a/include/sys/zfs_context.h
+++ b/include/sys/zfs_context.h
@@ -382,7 +382,6 @@ extern kstat_t *kstat_create(char *, int,
 #define	TASKQ_DYNAMIC		0x0004	/* Use dynamic thread scheduling */
 #define	TASKQ_THREADS_CPU_PCT	0x0008	/* Scale # threads by # cpus */
 #define	TASKQ_DC_BATCH		0x0010	/* Mark threads as batch */
-#define	TASKQ_NORECLAIM		0x0020	/* Disable direct memory reclaim */
 
 #define	TQ_SLEEP	KM_SLEEP	/* Can block for memory */
 #define	TQ_NOSLEEP	KM_NOSLEEP	/* cannot block for memory; may fail */
diff --git a/module/zfs/spa.c b/module/zfs/spa.c
index c33a33a..436cd26 100644
--- a/module/zfs/spa.c
+++ b/module/zfs/spa.c
@@ -617,8 +617,9 @@ static inline int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config,
 
 static taskq_t *
 spa_taskq_create(spa_t *spa, const char *name, enum zti_modes mode,
-    uint_t value, uint_t flags)
+    uint_t value)
 {
+	uint_t flags = TASKQ_PREPOPULATE;
 	boolean_t batch = B_FALSE;
 
 	switch (mode) {
@@ -668,17 +669,13 @@ static inline int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config,
 			const zio_taskq_info_t *ztip = &zio_taskqs[t][q];
 			enum zti_modes mode = ztip->zti_mode;
 			uint_t value = ztip->zti_value;
-			uint_t flags = 0;
 			char name[32];
 
-			if (t == ZIO_TYPE_WRITE)
-				flags |= TASKQ_NORECLAIM;
-
 			(void) snprintf(name, sizeof (name),
 			    "%s_%s", zio_type_name[t], zio_taskq_types[q]);
 
 			spa->spa_zio_taskq[t][q] =
-			    spa_taskq_create(spa, name, mode, value, flags);
+			    spa_taskq_create(spa, name, mode, value);
 		}
 	}
 }
-- 
1.7.10




1.1                  sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-1-of-3.patch

file : http://sources.gentoo.org/viewvc.cgi/gentoo-x86/sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-1-of-3.patch?rev=1.1&view=markup
plain: http://sources.gentoo.org/viewvc.cgi/gentoo-x86/sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-1-of-3.patch?rev=1.1&content-type=text/plain

Index: zfs-kmod-0.6.0_rc9-remove-pfmalloc-1-of-3.patch
===================================================================
From e2060f922f8526aecf437de4cc4dff4c3cdfae16 Mon Sep 17 00:00:00 2001
From: Richard Yao <ryao@×××××××××××××.edu>
Date: Tue, 15 May 2012 23:19:32 -0400
Subject: [PATCH] Revert Fix ASSERTION(!dsl_pool_sync_context(tx->tx_pool))

Commit eec8164771bee067c3cd55ed0a16dadeeba276de worked around an issue
involving direct reclaim through the use of PF_MEMALLOC.   Since we
are reworking thing to use KM_PUSHPAGE so that swap works, we revert
this patch in favor of the use of KM_PUSHPAGE in the affected areas.

Signed-off-by: Richard Yao <ryao@×××××××××××××.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@××××.gov>
Issue #726
---
 module/zfs/txg.c |   13 -------------
 1 file changed, 13 deletions(-)

diff --git a/module/zfs/txg.c b/module/zfs/txg.c
index 5021e44..8d037f7 100644
--- a/module/zfs/txg.c
+++ b/module/zfs/txg.c
@@ -372,19 +372,6 @@
 	callb_cpr_t cpr;
 	uint64_t start, delta;
 
-#ifdef _KERNEL
-	/*
-	 * Disable the normal reclaim path for the txg_sync thread.  This
-	 * ensures the thread will never enter dmu_tx_assign() which can
-	 * otherwise occur due to direct reclaim.  If this is allowed to
-	 * happen the system can deadlock.  Direct reclaim call path:
-	 *
-	 *   ->shrink_icache_memory->prune_icache->dispose_list->
-	 *   clear_inode->zpl_clear_inode->zfs_inactive->dmu_tx_assign
-	 */
-	current->flags |= PF_MEMALLOC;
-#endif /* _KERNEL */
-
 	txg_thread_enter(tx, &cpr);
 
 	start = delta = 0;
-- 
1.7.10




1.1                  sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-2-of-3.patch

file : http://sources.gentoo.org/viewvc.cgi/gentoo-x86/sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-2-of-3.patch?rev=1.1&view=markup
plain: http://sources.gentoo.org/viewvc.cgi/gentoo-x86/sys-fs/zfs-kmod/files/zfs-kmod-0.6.0_rc9-remove-pfmalloc-2-of-3.patch?rev=1.1&content-type=text/plain

Index: zfs-kmod-0.6.0_rc9-remove-pfmalloc-2-of-3.patch
===================================================================
From 78d26b8497b3845fc8130981c76214d6788f7a9d Mon Sep 17 00:00:00 2001
From: Richard Yao <ryao@×××××××××××××.edu>
Date: Mon, 7 May 2012 14:14:45 -0400
Subject: [PATCH] Revert Fix zpl_writepage() deadlock

The commit, cfc9a5c88f91f7b4d606fce89505e1f404691ea5, to fix deadlocks
in zpl_writepage() relied on PF_MEMALLOC.   That had the effect of
disabling the direct reclaim path on all allocations originating from
calls to this function, but it failed to address the actual cause of
those deadlocks.  This led to the same deadlocks being observed with
swap on zvols, but not with swap on the loop device, which exercises
this code.

The use of PF_MEMALLOC also had the side effect of permitting
allocations to be made from ZONE_DMA in instances that did not require
it.  This contributes to the possibility of panics caused by depletion
of pages from ZONE_DMA.

As such, we revert this patch in favor of a proper fix for both issues.

Signed-off-by: Richard Yao <ryao@×××××××××××××.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@××××.gov>
Issue #726
---
 module/zfs/zpl_file.c |   15 +--------------
 1 file changed, 1 insertion(+), 14 deletions(-)

diff --git a/module/zfs/zpl_file.c b/module/zfs/zpl_file.c
index 5ac41c9..2e9f72a 100644
--- a/module/zfs/zpl_file.c
+++ b/module/zfs/zpl_file.c
@@ -358,20 +358,7 @@
 	ASSERT(PageLocked(pp));
 	ASSERT(!PageWriteback(pp));
 
-	/*
-	 * Disable the normal reclaim path for zpl_putpage().  This
-	 * ensures that all memory allocations under this call path
-	 * will never enter direct reclaim.  If this were to happen
-	 * the VM might try to write out additional pages by calling
-	 * zpl_putpage() again resulting in a deadlock.
-	 */
-	if (current->flags & PF_MEMALLOC) {
-		(void) zfs_putpage(mapping->host, pp, wbc);
-	} else {
-		current->flags |= PF_MEMALLOC;
-		(void) zfs_putpage(mapping->host, pp, wbc);
-		current->flags &= ~PF_MEMALLOC;
-	}
+	(void) zfs_putpage(mapping->host, pp, wbc);
 
 	return (0);
 }
-- 
1.7.10