[Cluster-devel] GFS2: Pre-pull patch posting (fixes)

[Cluster-devel] GFS2: Pre-pull patch posting (fixes)

[Steven Whitehouse]

There are four patches this time. The first fixes a problem where the
wrong descriptor type was being written into the log for journaled data
blocks. The second fixes a race relating to the deallocation of allocator
data. The third provides a fallback if kmalloc is unable to satisfy a
request to allocate a directory hash table. The fourth fixes the iopen
glock caching so that inodes are deleted in a more timely manner after
rmdir/unlink,

Steve.

[Cluster-devel] [PATCH 1/4] GFS2: Set log descriptor type for jdata blocks

[Steven Whitehouse]

From: Bob Peterson <rpeterso@redhat.com>

This patch sets the log descriptor type according to whether the
journal commit is for (journaled) data or metadata. This was
recently broken when the functions to process data and metadata
log ops were combined.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/lops.c b/fs/gfs2/lops.c
index 68b4c8f..6c33d7b 100644
--- a/fs/gfs2/lops.c
+++ b/fs/gfs2/lops.c
@@ -419,7 +419,9 @@ static void gfs2_before_commit(struct gfs2_sbd *sdp, unsigned int limit,
 		if (total > limit)
 			num = limit;
 		gfs2_log_unlock(sdp);
-		page = gfs2_get_log_desc(sdp, GFS2_LOG_DESC_METADATA, num + 1, num);
+		page = gfs2_get_log_desc(sdp,
+					 is_databuf ? GFS2_LOG_DESC_JDATA :
+					 GFS2_LOG_DESC_METADATA, num + 1, num);
 		ld = page_address(page);
 		gfs2_log_lock(sdp);
 		ptr = (__be64 *)(ld + 1);
-- 
1.7.4

[Cluster-devel] [PATCH 2/4] GFS2: Increase i_writecount during gfs2_setattr_size

[Steven Whitehouse]

From: Bob Peterson <rpeterso@redhat.com>

This patch calls get_write_access in a few functions. This
merely increases inode->i_writecount for the duration of the function.
That will ensure that any file closes won't delete the inode's
multi-block reservation while the function is running.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
index 1dc9a13..93b5809 100644
--- a/fs/gfs2/bmap.c
+++ b/fs/gfs2/bmap.c
@@ -1286,17 +1286,26 @@ int gfs2_setattr_size(struct inode *inode, u64 newsize)
 	if (ret)
 		return ret;
 
+	ret = get_write_access(inode);
+	if (ret)
+		return ret;
+
 	inode_dio_wait(inode);
 
 	ret = gfs2_rs_alloc(GFS2_I(inode));
 	if (ret)
-		return ret;
+		goto out;
 
 	oldsize = inode->i_size;
-	if (newsize >= oldsize)
-		return do_grow(inode, newsize);
+	if (newsize >= oldsize) {
+		ret = do_grow(inode, newsize);
+		goto out;
+	}
 
-	return do_shrink(inode, oldsize, newsize);
+	ret = do_shrink(inode, oldsize, newsize);
+out:
+	put_write_access(inode);
+	return ret;
 }
 
 int gfs2_truncatei_resume(struct gfs2_inode *ip)
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index acd1676..ad0dc38 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -402,16 +402,20 @@ static int gfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	/* Update file times before taking page lock */
 	file_update_time(vma->vm_file);
 
+	ret = get_write_access(inode);
+	if (ret)
+		goto out;
+
 	ret = gfs2_rs_alloc(ip);
 	if (ret)
-		return ret;
+		goto out_write_access;
 
 	gfs2_size_hint(vma->vm_file, pos, PAGE_CACHE_SIZE);
 
 	gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
 	ret = gfs2_glock_nq(&gh);
 	if (ret)
-		goto out;
+		goto out_uninit;
 
 	set_bit(GLF_DIRTY, &ip->i_gl->gl_flags);
 	set_bit(GIF_SW_PAGED, &ip->i_flags);
@@ -480,12 +484,15 @@ out_quota_unlock:
 	gfs2_quota_unlock(ip);
 out_unlock:
 	gfs2_glock_dq(&gh);
-out:
+out_uninit:
 	gfs2_holder_uninit(&gh);
 	if (ret == 0) {
 		set_page_dirty(page);
 		wait_for_stable_page(page);
 	}
+out_write_access:
+	put_write_access(inode);
+out:
 	sb_end_pagefault(inode->i_sb);
 	return block_page_mkwrite_return(ret);
 }
@@ -594,10 +601,10 @@ static int gfs2_release(struct inode *inode, struct file *file)
 	kfree(file->private_data);
 	file->private_data = NULL;
 
-	if ((file->f_mode & FMODE_WRITE) &&
-	    (atomic_read(&inode->i_writecount) == 1))
-		gfs2_rs_delete(ip);
+	if (!(file->f_mode & FMODE_WRITE))
+		return 0;
 
+	gfs2_rs_delete(ip);
 	return 0;
 }
 
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index 5232525..9809156 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -638,8 +638,10 @@ void gfs2_rs_deltree(struct gfs2_blkreserv *rs)
  */
 void gfs2_rs_delete(struct gfs2_inode *ip)
 {
+	struct inode *inode = &ip->i_inode;
+
 	down_write(&ip->i_rw_mutex);
-	if (ip->i_res) {
+	if (ip->i_res && atomic_read(&inode->i_writecount) <= 1) {
 		gfs2_rs_deltree(ip->i_res);
 		BUG_ON(ip->i_res->rs_free);
 		kmem_cache_free(gfs2_rsrv_cachep, ip->i_res);
-- 
1.7.4

[Cluster-devel] [PATCH 3/4] GFS2: Fall back to vmalloc if kmalloc fails for dir hash tables

[Steven Whitehouse]

From: Bob Peterson <rpeterso@redhat.com>

This version has one more correction: the vmalloc calls are replaced
by __vmalloc calls to preserve the GFP_NOFS flag.

When GFS2's directory management code allocates buffers for a
directory hash table, if it can't get the memory it needs, it
currently gives a bad return code. Rather than giving an error,
this patch allows it to use virtual memory rather than kernel
memory for the hash table. This should make it possible for
directories to function properly, even when kernel memory becomes
very fragmented.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/dir.c b/fs/gfs2/dir.c
index c3e82bd..b631c90 100644
--- a/fs/gfs2/dir.c
+++ b/fs/gfs2/dir.c
@@ -354,22 +354,31 @@ static __be64 *gfs2_dir_get_hash_table(struct gfs2_inode *ip)
 		return ERR_PTR(-EIO);
 	}
 
-	hc = kmalloc(hsize, GFP_NOFS);
-	ret = -ENOMEM;
+	hc = kmalloc(hsize, GFP_NOFS | __GFP_NOWARN);
+	if (hc == NULL)
+		hc = __vmalloc(hsize, GFP_NOFS, PAGE_KERNEL);
+
 	if (hc == NULL)
 		return ERR_PTR(-ENOMEM);
 
 	ret = gfs2_dir_read_data(ip, hc, hsize);
 	if (ret < 0) {
-		kfree(hc);
+		if (is_vmalloc_addr(hc))
+			vfree(hc);
+		else
+			kfree(hc);
 		return ERR_PTR(ret);
 	}
 
 	spin_lock(&inode->i_lock);
-	if (ip->i_hash_cache)
-		kfree(hc);
-	else
+	if (ip->i_hash_cache) {
+		if (is_vmalloc_addr(hc))
+			vfree(hc);
+		else
+			kfree(hc);
+	} else {
 		ip->i_hash_cache = hc;
+	}
 	spin_unlock(&inode->i_lock);
 
 	return ip->i_hash_cache;
@@ -385,7 +394,10 @@ void gfs2_dir_hash_inval(struct gfs2_inode *ip)
 {
 	__be64 *hc = ip->i_hash_cache;
 	ip->i_hash_cache = NULL;
-	kfree(hc);
+	if (is_vmalloc_addr(hc))
+		vfree(hc);
+	else
+		kfree(hc);
 }
 
 static inline int gfs2_dirent_sentinel(const struct gfs2_dirent *dent)
@@ -1113,7 +1125,10 @@ static int dir_double_exhash(struct gfs2_inode *dip)
 	if (IS_ERR(hc))
 		return PTR_ERR(hc);
 
-	h = hc2 = kmalloc(hsize_bytes * 2, GFP_NOFS);
+	h = hc2 = kmalloc(hsize_bytes * 2, GFP_NOFS | __GFP_NOWARN);
+	if (hc2 == NULL)
+		hc2 = __vmalloc(hsize_bytes * 2, GFP_NOFS, PAGE_KERNEL);
+
 	if (!hc2)
 		return -ENOMEM;
 
@@ -1145,7 +1160,10 @@ fail:
 	gfs2_dinode_out(dip, dibh->b_data);
 	brelse(dibh);
 out_kfree:
-	kfree(hc2);
+	if (is_vmalloc_addr(hc2))
+		vfree(hc2);
+	else
+		kfree(hc2);
 	return error;
 }
 
@@ -1846,6 +1864,8 @@ static int leaf_dealloc(struct gfs2_inode *dip, u32 index, u32 len,
 	memset(&rlist, 0, sizeof(struct gfs2_rgrp_list));
 
 	ht = kzalloc(size, GFP_NOFS);
+	if (ht == NULL)
+		ht = vzalloc(size);
 	if (!ht)
 		return -ENOMEM;
 
@@ -1933,7 +1953,10 @@ out_rlist:
 	gfs2_rlist_free(&rlist);
 	gfs2_quota_unhold(dip);
 out:
-	kfree(ht);
+	if (is_vmalloc_addr(ht))
+		vfree(ht);
+	else
+		kfree(ht);
 	return error;
 }
 
-- 
1.7.4

[Cluster-devel] [PATCH 4/4] GFS2: Don't cache iopen glocks

[Steven Whitehouse]

From: Bob Peterson <rpeterso@redhat.com>

This patch makes GFS2 immediately reclaim/delete all iopen glocks
as soon as they're dequeued. This allows deleters to get an
EXclusive lock on iopen so files are deleted properly instead of
being set as unlinked.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 8833a4f..62b484e 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -189,6 +189,7 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type,
 	return inode;
 
 fail_refresh:
+	ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 	ip->i_iopen_gh.gh_gl->gl_object = NULL;
 	gfs2_glock_dq_uninit(&ip->i_iopen_gh);
 fail_iopen:
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 917c8e1..e5639de 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -1444,6 +1444,7 @@ static void gfs2_evict_inode(struct inode *inode)
 	/* Must not read inode block until block type has been verified */
 	error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, GL_SKIP, &gh);
 	if (unlikely(error)) {
+		ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 		gfs2_glock_dq_uninit(&ip->i_iopen_gh);
 		goto out;
 	}
@@ -1514,8 +1515,10 @@ out_unlock:
 	if (gfs2_rs_active(ip->i_res))
 		gfs2_rs_deltree(ip->i_res);
 
-	if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags))
+	if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) {
+		ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 		gfs2_glock_dq(&ip->i_iopen_gh);
+	}
 	gfs2_holder_uninit(&ip->i_iopen_gh);
 	gfs2_glock_dq_uninit(&gh);
 	if (error && error != GLR_TRYFAILED && error != -EROFS)
@@ -1534,6 +1537,7 @@ out:
 	ip->i_gl = NULL;
 	if (ip->i_iopen_gh.gh_gl) {
 		ip->i_iopen_gh.gh_gl->gl_object = NULL;
+		ip->i_iopen_gh.gh_flags |= GL_NOCACHE;
 		gfs2_glock_dq_uninit(&ip->i_iopen_gh);
 	}
 }
-- 
1.7.4