Commit fe0bbd29 authored Jun 23, 2014 by Steven Whitehouse

GFS2: Use GFP_NOFS when allocating glocks



Normally GFP_KERNEL is ok here, but there is now a rarely used code path
relating to deallocation of unlinked inodes (in certain corner cases)
which if hit at times of memory shortage can cause recursion while
trying to free memory.

One solution would be to try and move the gfs2_glock_get() call so
that it is no longer called while another glock is held, but that
doesn't look at all easy, so GFP_NOFS is the best solution for the
time being.

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

parent 94a09a39

fs/gfs2/glock.c

+2 −2

Original line number	Diff line number	Diff line
		@@ -731,14 +731,14 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
		cachep = gfs2_glock_aspace_cachep;
		else
		cachep = gfs2_glock_cachep;
		gl = kmem_cache_alloc(cachep, GFP_KERNEL);
		gl = kmem_cache_alloc(cachep, GFP_NOFS);
		if (!gl)
		return -ENOMEM;

		memset(&gl->gl_lksb, 0, sizeof(struct dlm_lksb));

		if (glops->go_flags & GLOF_LVB) {
		gl->gl_lksb.sb_lvbptr = kzalloc(GFS2_MIN_LVB_SIZE, GFP_KERNEL);
		gl->gl_lksb.sb_lvbptr = kzalloc(GFS2_MIN_LVB_SIZE, GFP_NOFS);
		if (!gl->gl_lksb.sb_lvbptr) {
		kmem_cache_free(cachep, gl);
		return -ENOMEM;