GFS2: Fix rgrp end rounding problem for bsize < page size
authorBob Peterson <rpeterso@redhat.com>
Wed, 28 Oct 2015 14:05:31 +0000 (09:05 -0500)
committerBob Peterson <rpeterso@redhat.com>
Mon, 9 Nov 2015 15:38:02 +0000 (09:38 -0600)
This patch fixes a bug introduced by commit 7005c3e. That patch
tries to map a vm range for resource groups, but the calculation
breaks down when the block size is less than the page size.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
fs/gfs2/rgrp.c

index d29dd0cec9140fa10d62eecb549d5d206b0fd0e6..c134c0462ceecfaa9df64d916d0ed6762b937c18 100644 (file)
@@ -933,8 +933,9 @@ static int read_rindex_entry(struct gfs2_inode *ip)
                goto fail;
 
        rgd->rd_gl->gl_object = rgd;
-       rgd->rd_gl->gl_vm.start = rgd->rd_addr * bsize;
-       rgd->rd_gl->gl_vm.end = rgd->rd_gl->gl_vm.start + (rgd->rd_length * bsize) - 1;
+       rgd->rd_gl->gl_vm.start = (rgd->rd_addr * bsize) & PAGE_CACHE_MASK;
+       rgd->rd_gl->gl_vm.end = PAGE_CACHE_ALIGN((rgd->rd_addr +
+                                                 rgd->rd_length) * bsize) - 1;
        rgd->rd_rgl = (struct gfs2_rgrp_lvb *)rgd->rd_gl->gl_lksb.sb_lvbptr;
        rgd->rd_flags &= ~(GFS2_RDF_UPTODATE | GFS2_RDF_PREFERRED);
        if (rgd->rd_data > sdp->sd_max_rg_data)