commit d033351753e1c45e6be74342d05de0a2501d5211

Author: Abhi Das <adas@redhat.com>

Date:   Mon May 11 09:22:31 2020 -0500

    gfs2_jadd: Handle out-of-space issues

    If gfs2_jadd runs out of disk space while adding journals, it does

    not exit gracefully. It partially does its job and bails out when

    it hits -ENOSPC. This leaves the metafs mounted and most likely a

    corrupted filesystem that even fsck.gfs2 can't fix.

    This patch adds a pre-check that ensures that the journals requested

    will fit in the available space before proceeding. Note that this is

    not foolproof because gfs2_jadd operates on a mounted filesystem.

    While it is required that the filesystem be idle (and mounted on only

    one node) while gfs2_jadd is being run, there is nothing stopping a

    user from having some I/O process competing with gfs2_jadd for disk

    blocks and consequently crashing it.

    Resolves: rhbz#1833141

    Signed-off-by: Abhi Das <adas@redhat.com>

diff --git a/gfs2/mkfs/main_jadd.c b/gfs2/mkfs/main_jadd.c

index efe91e30..c5424803 100644

--- a/gfs2/mkfs/main_jadd.c

+++ b/gfs2/mkfs/main_jadd.c

@@ -396,6 +396,8 @@ static void gather_info(struct gfs2_sbd *sdp, struct jadd_opts *opts)

 		exit(EXIT_FAILURE);

 	}

 	sdp->bsize = statbuf.f_bsize;

+	sdp->blks_total = statbuf.f_blocks;

+	sdp->blks_alloced = sdp->blks_total - statbuf.f_bfree;

 }

 static void find_current_journals(struct jadd_opts *opts)

@@ -527,13 +529,43 @@ static void add_j(struct gfs2_sbd *sdp, struct jadd_opts *opts)

 	}

 }

+static int check_fit(struct gfs2_sbd *sdp, struct jadd_opts *opts)

+{

+	/* Compute how much space we'll need for the new journals

+	 * Number of blocks needed per added journal:

+	 * 1 block for the ir inode

+	 * 1 block for the sc inode

+	 * for sizes of the qc and journal inodes, use lgfs2_space_for_data()

+	 * to calculate.

+	 */

+	uint64_t blks_per_j, total_blks;

+

+	blks_per_j = 1 + 1 +

+		lgfs2_space_for_data(sdp, sdp->bsize, sdp->qcsize << 20) +

+		lgfs2_space_for_data(sdp, sdp->bsize, sdp->jsize << 20);

+	total_blks = opts->journals * blks_per_j;

+

+	if (total_blks > (sdp->blks_total - sdp->blks_alloced)) {

+		printf( _("\nInsufficient space on the device to add %u %uMB "

+			  "journals (%uMB QC size)\n\n"),

+			opts->journals, sdp->jsize, sdp->qcsize);

+		printf( _("Required space  : %*lu blks (%lu blks per "

+			  "journal)\n"), 10, total_blks, blks_per_j);

+		printf( _("Available space : %*lu blks\n\n"), 10,

+			sdp->blks_total - sdp->blks_alloced);

+		errno = ENOSPC;

+		return 1;

+	}

+	return 0;

+}

+

 int main(int argc, char *argv[])

 {

 	struct jadd_opts opts = {0};

 	struct gfs2_sbd sbd, *sdp = &sbd;

 	struct metafs mfs = {0};

 	struct mntent *mnt;

-	unsigned int total;

+	unsigned int total, ret = 0;

 	setlocale(LC_ALL, "");

 	textdomain("gfs2-utils");

@@ -574,6 +606,12 @@ int main(int argc, char *argv[])

 	}

 	find_current_journals(&opts);

+	ret = check_fit(sdp, &opts);

+	if (ret) {

+		perror(_("Failed to add journals"));

+		goto out;

+	}

+

 	total = opts.orig_journals + opts.journals;

 	for (opts.journals = opts.orig_journals;

 	     opts.journals < total;

@@ -588,13 +626,16 @@ int main(int argc, char *argv[])

 		add_j(sdp, &opts);

 	}

+out:

 	free(opts.new_inode);

 	free(opts.per_node);

 	free(opts.jindex);

 	close(sdp->path_fd);

 	cleanup_metafs(&mfs;;

 	sync();

-	print_results(&opts);

-	return 0;

+	if (!ret)

+		print_results(&opts);

+

+	return ret;

 }