cluster-commits June 2011

cluster-commits@lists.stg.fedorahosted.org

7 participants
32 discussions

cluster: the annotated tag cluster-3.1.3 has been created
by Fabio M. Di Nitto 20 Jun '11

20 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=7e… Commit: 7eb116df7cfad2970ce4db22b34ba6ba01ee4948 Parent: 0000000000000000000000000000000000000000 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: 2011-06-20 15:40 +0000 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: 2011-06-20 15:40 +0000 annotated tag: cluster-3.1.3 has been created at 7eb116df7cfad2970ce4db22b34ba6ba01ee4948 (tag) tagging ef1c1dd818cf38d27455ce0e4678d55ae5c499eb (commit) replaces cluster-3.1.2 cluster-3.1.3 release Fabio M. Di Nitto (2): cman: fix ttl default if no value is specified dlm_controld: fix build with old (< 3.0) kernel headers

1 0

cluster: STABLE31 - dlm_controld: fix build with old (< 3.0) kernel headers
by Fabio M. Di Nitto 20 Jun '11

20 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=ef… Commit: ef1c1dd818cf38d27455ce0e4678d55ae5c499eb Parent: f0b85f3680510b03058a1e6e94fc11ad913ff59b Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Mon Jun 20 17:34:43 2011 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Mon Jun 20 17:35:59 2011 +0200 dlm_controld: fix build with old (< 3.0) kernel headers Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- group/dlm_controld/plock.c | 13 +++++++++---- 1 files changed, 9 insertions(+), 4 deletions(-) diff --git a/group/dlm_controld/plock.c b/group/dlm_controld/plock.c index f596d2f..166f6c9 100644 --- a/group/dlm_controld/plock.c +++ b/group/dlm_controld/plock.c @@ -3,10 +3,6 @@ #include <linux/dlm_plock.h> -/* FIXME: remove this once everyone is using the version of - * dlm_plock.h which defines it */ -#define DLM_PLOCK_FL_CLOSE 1 - static uint32_t plock_read_count; static uint32_t plock_recv_count; static uint32_t plock_rate_delays; @@ -741,7 +737,16 @@ static void do_unlock(struct lockspace *ls, struct dlm_plock_info *in, rv = unlock_internal(ls, r, in); +/* FIXME: remove this once everyone is using the version of + * dlm_plock.h which defines it */ + +#ifndef DLM_PLOCK_FL_CLOSE +#warning DLM_PLOCK_FL_CLOSE undefined. Enabling build workaround. +#define DLM_PLOCK_FL_CLOSE 1 + if (in->pad & DLM_PLOCK_FL_CLOSE) { +#else if (in->flags & DLM_PLOCK_FL_CLOSE) { +#endif clear_waiters(ls, r, in); /* no replies for unlock-close ops */ goto skip_result;

1 0

dlm: master - dlm_controld: fix build with old (< 3.0) kernel headers
by Fabio M. Di Nitto 20 Jun '11

20 Jun '11

Gitweb: http://git.fedorahosted.org/git/dlm.git?p=dlm.git;a=commitdiff;h=13a8144335… Commit: 13a814433588ba0a9d8228dcb2e6c6dba6f3c861 Parent: 1e95a3d8c8fb8ef889b4e69fca9ac7d9c062517b Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Mon Jun 20 17:34:43 2011 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Mon Jun 20 17:34:43 2011 +0200 dlm_controld: fix build with old (< 3.0) kernel headers Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- group/dlm_controld/plock.c | 13 +++++++++---- 1 files changed, 9 insertions(+), 4 deletions(-) diff --git a/group/dlm_controld/plock.c b/group/dlm_controld/plock.c index 967f3c0..3848308 100644 --- a/group/dlm_controld/plock.c +++ b/group/dlm_controld/plock.c @@ -3,10 +3,6 @@ #include <linux/dlm_plock.h> -/* FIXME: remove this once everyone is using the version of - * dlm_plock.h which defines it */ -#define DLM_PLOCK_FL_CLOSE 1 - static uint32_t plock_read_count; static uint32_t plock_recv_count; static uint32_t plock_rate_delays; @@ -789,7 +785,16 @@ static void do_unlock(struct lockspace *ls, struct dlm_plock_info *in, rv = unlock_internal(ls, r, in); +/* FIXME: remove this once everyone is using the version of + * dlm_plock.h which defines it */ + +#ifndef DLM_PLOCK_FL_CLOSE +#warning DLM_PLOCK_FL_CLOSE undefined. Enabling build workaround. +#define DLM_PLOCK_FL_CLOSE 1 + if (in->pad & DLM_PLOCK_FL_CLOSE) { +#else if (in->flags & DLM_PLOCK_FL_CLOSE) { +#endif clear_waiters(ls, r, in); /* no replies for unlock-close ops */ goto skip_result;

1 0

cluster: RHEL6 - cman: fix ttl default if no value is specified
by Fabio M. Di Nitto 20 Jun '11

20 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=8a… Commit: 8aed0b01c8372bde746132c6aea3465a12cd164d Parent: 0b505a0b890dfe8898a612cd45667f0ba2086304 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Mon Jun 20 09:49:34 2011 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Mon Jun 20 15:33:48 2011 +0200 cman: fix ttl default if no value is specified when specifing a multicast address, ttl was set to 0 unless ttl was explicitly set in cluster.conf. Resolves: rhbz#713977 Reviewed-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/daemon/cman-preconfig.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index 5a73fb3..cde130b 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -691,7 +691,7 @@ static int get_nodename(struct objdb_iface_ver0 *objdb) if (objdb->object_find_next(find_handle2, &mcast_handle) == 0) { objdb_get_string(objdb, mcast_handle, "addr", &mcast_name); - objdb_get_int(objdb, mcast_handle, "ttl", &ttl, 0); + objdb_get_int(objdb, mcast_handle, "ttl", &ttl, ttl); } objdb->object_find_destroy(find_handle2); }

1 0

cluster: STABLE31 - cman: fix ttl default if no value is specified
by Fabio M. Di Nitto 20 Jun '11

20 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=f0… Commit: f0b85f3680510b03058a1e6e94fc11ad913ff59b Parent: a548bd3333824341373a5f9a8ff9a456c5af59c8 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Mon Jun 20 09:49:34 2011 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Mon Jun 20 09:49:34 2011 +0200 cman: fix ttl default if no value is specified when specifing a multicast address, ttl was set to 0 unless ttl was explicitly set in cluster.conf. Resolves: rhbz#713977 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/daemon/cman-preconfig.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/cman/daemon/cman-preconfig.c b/cman/daemon/cman-preconfig.c index a890fb3..aeedf8b 100644 --- a/cman/daemon/cman-preconfig.c +++ b/cman/daemon/cman-preconfig.c @@ -673,7 +673,7 @@ static int get_nodename(struct objdb_iface_ver0 *objdb) if (objdb->object_find_next(find_handle2, &mcast_handle) == 0) { objdb_get_string(objdb, mcast_handle, "addr", &mcast_name); - objdb_get_int(objdb, mcast_handle, "ttl", &ttl, 0); + objdb_get_int(objdb, mcast_handle, "ttl", &ttl, ttl); } objdb->object_find_destroy(find_handle2); }

1 0

cluster: RHEL6 - fsck.gfs2 only rebuilds one missing journal at a time
by Bob Peterson 16 Jun '11

16 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=0b… Commit: 0b505a0b890dfe8898a612cd45667f0ba2086304 Parent: d02df4bcd1aee3ed287ee8f8c08c34c47db4cd1e Author: Bob Peterson <rpeterso(a)redhat.com> AuthorDate: Thu Jun 16 13:50:54 2011 -0500 Committer: Bob Peterson <rpeterso(a)redhat.com> CommitterDate: Thu Jun 16 13:50:54 2011 -0500 fsck.gfs2 only rebuilds one missing journal at a time This patch allows fsck.gfs2 to rebuild multiple journals in a single run. It figures out the correct number of journals based on the number of entries in the per_node directory. If per_node is missing, it reverts to its previous behavior of rebuilding a single journal. rhbz#683104 --- gfs2/fsck/fs_recovery.c | 16 +++++++++--- gfs2/fsck/initialize.c | 62 ++++++++++++++++++++++++++++++++++++++++------ 2 files changed, 66 insertions(+), 12 deletions(-) diff --git a/gfs2/fsck/fs_recovery.c b/gfs2/fsck/fs_recovery.c index 07c1552..b9d856e 100644 --- a/gfs2/fsck/fs_recovery.c +++ b/gfs2/fsck/fs_recovery.c @@ -636,20 +636,28 @@ int ji_update(struct gfs2_sbd *sdp) return -1; } - if(!(sdp->md.journal = calloc(ip->i_di.di_entries - 2, sizeof(struct gfs2_inode *)))) { + /* The per_node directory will have 3 directory entries per node, + plus two for "." and "..". So we subtract the 2 and divide by 3. + If per_node is missing or damaged, we have to trust jindex has + the correct number of entries. */ + if (sdp->md.pinode) /* if per_node was read in properly */ + sdp->md.journals = (sdp->md.pinode->i_di.di_entries - 2) / 3; + else + sdp->md.journals = ip->i_di.di_entries - 2; + + if(!(sdp->md.journal = calloc(sdp->md.journals, + sizeof(struct gfs2_inode *)))) { log_err("Unable to allocate journal index\n"); return -1; } - sdp->md.journals = 0; memset(journal_name, 0, sizeof(*journal_name)); - for(i = 0; i < ip->i_di.di_entries - 2; i++) { + for (i = 0; i < sdp->md.journals; i++) { /* FIXME check snprintf return code */ snprintf(journal_name, JOURNAL_NAME_SIZE, "journal%u", i); gfs2_lookupi(sdp->md.jiinode, journal_name, strlen(journal_name), &jip); sdp->md.journal[i] = jip; } - sdp->md.journals = ip->i_di.di_entries - 2; return 0; } diff --git a/gfs2/fsck/initialize.c b/gfs2/fsck/initialize.c index 469f022..5a6fc6e 100644 --- a/gfs2/fsck/initialize.c +++ b/gfs2/fsck/initialize.c @@ -391,6 +391,52 @@ static int rebuild_master(struct gfs2_sbd *sdp) } /** + * lookup_per_node - Make sure the per_node directory is read in + * + * This function is used to read in the per_node directory. It is called + * twice. The first call tries to read in the dinode early on. That ensures + * that if any journals are missing, we can figure out the number of journals + * from per_node. However, we unfortunately can't rebuild per_node at that + * point in time because our resource groups aren't read in yet. + * The second time it's called is much later when we can rebuild it. + * + * allow_rebuild: 0 if rebuilds are not allowed + * 1 if rebuilds are allowed + */ +static void lookup_per_node(struct gfs2_sbd *sdp, int allow_rebuild) +{ + if (sdp->md.pinode) + return; + + gfs2_lookupi(sdp->master_dir, "per_node", 8, &sdp->md.pinode); + if (sdp->md.pinode) + return; + if (!allow_rebuild) { + log_err( _("The gfs2 system per_node directory " + "inode is missing, so we might not be \nable to " + "rebuild missing journals this run.\n")); + return; + } + + if (query( _("The gfs2 system per_node directory " + "inode is missing. Okay to rebuild it? (y/n) "))) { + int err; + + err = build_per_node(sdp); + if (err) { + log_crit(_("Error rebuilding per_node directory: %s\n"), + strerror(err)); + exit(-1); + } + } + gfs2_lookupi(sdp->master_dir, "per_node", 8, &sdp->md.pinode); + if (!sdp->md.pinode) { + log_err( _("Unable to rebuild per_node; aborting.\n")); + exit(-1); + } +} + +/** * init_system_inodes * * Returns: 0 on success, -1 on failure @@ -520,14 +566,9 @@ static int init_system_inodes(struct gfs2_sbd *sdp) build_quota(sdp); } - gfs2_lookupi(sdp->master_dir, "per_node", 8, &sdp->md.pinode); - if (!sdp->md.pinode) { - if (query( _("The gfs2 system per_node directory inode is " - "missing. Okay to rebuild it? (y/n) "))) - build_per_node(sdp); - } - - /* FIXME fill in per_node structure */ + /* Try to lookup the per_node inode. If it was missing, it is now + safe to rebuild it. */ + lookup_per_node(sdp, 1); /******************************************************************* ******* Now, set boundary fields in the super block ************* *******************************************************************/ @@ -1126,6 +1167,11 @@ int initialize(struct gfs2_sbd *sbp, int force_check, int preen, sbp->sd_sb.sb_master_dir.no_addr); } + /* Look up the "per_node" inode. If there are journals missing, we + need to figure out what's missing from per_node. And we need all + our journals to be there before we can replay them. */ + lookup_per_node(sbp, 0); + /* verify various things */ if(replay_journals(sbp, preen, force_check, &clean_journals)) {

1 0

gfs2-utils: master - fsck.gfs2 only rebuilds one missing journal at a time
by Bob Peterson 16 Jun '11

16 Jun '11

Gitweb: http://git.fedorahosted.org/git/gfs2-utils.git?p=gfs2-utils.git;a=commitdif… Commit: 96e3a7f031d04d792b1e24758849015942631593 Parent: 8331ddfd39725e0d206633488cf5582c252e9aa5 Author: Bob Peterson <rpeterso(a)redhat.com> AuthorDate: Thu Jun 16 13:37:10 2011 -0500 Committer: Bob Peterson <rpeterso(a)redhat.com> CommitterDate: Thu Jun 16 13:37:10 2011 -0500 fsck.gfs2 only rebuilds one missing journal at a time This patch allows fsck.gfs2 to rebuild multiple journals in a single run. It figures out the correct number of journals based on the number of entries in the per_node directory. If per_node is missing, it reverts to its previous behavior of rebuilding a single journal. rhbz#683104 --- gfs2/fsck/fs_recovery.c | 16 ++++++++--- gfs2/fsck/initialize.c | 68 +++++++++++++++++++++++++++++++++++++--------- 2 files changed, 66 insertions(+), 18 deletions(-) diff --git a/gfs2/fsck/fs_recovery.c b/gfs2/fsck/fs_recovery.c index 21f6db1..da06598 100644 --- a/gfs2/fsck/fs_recovery.c +++ b/gfs2/fsck/fs_recovery.c @@ -638,20 +638,28 @@ int ji_update(struct gfs2_sbd *sdp) return -1; } - if(!(sdp->md.journal = calloc(ip->i_di.di_entries - 2, sizeof(struct gfs2_inode *)))) { + /* The per_node directory will have 3 directory entries per node, + plus two for "." and "..". So we subtract the 2 and divide by 3. + If per_node is missing or damaged, we have to trust jindex has + the correct number of entries. */ + if (sdp->md.pinode) /* if per_node was read in properly */ + sdp->md.journals = (sdp->md.pinode->i_di.di_entries - 2) / 3; + else + sdp->md.journals = ip->i_di.di_entries - 2; + + if(!(sdp->md.journal = calloc(sdp->md.journals, + sizeof(struct gfs2_inode *)))) { log_err("Unable to allocate journal index\n"); return -1; } - sdp->md.journals = 0; memset(journal_name, 0, sizeof(*journal_name)); - for(i = 0; i < ip->i_di.di_entries - 2; i++) { + for (i = 0; i < sdp->md.journals; i++) { /* FIXME check snprintf return code */ snprintf(journal_name, JOURNAL_NAME_SIZE, "journal%u", i); gfs2_lookupi(sdp->md.jiinode, journal_name, strlen(journal_name), &jip); sdp->md.journal[i] = jip; } - sdp->md.journals = ip->i_di.di_entries - 2; return 0; } diff --git a/gfs2/fsck/initialize.c b/gfs2/fsck/initialize.c index c5dc7ef..697671c 100644 --- a/gfs2/fsck/initialize.c +++ b/gfs2/fsck/initialize.c @@ -449,6 +449,52 @@ static int rebuild_master(struct gfs2_sbd *sdp) } /** + * lookup_per_node - Make sure the per_node directory is read in + * + * This function is used to read in the per_node directory. It is called + * twice. The first call tries to read in the dinode early on. That ensures + * that if any journals are missing, we can figure out the number of journals + * from per_node. However, we unfortunately can't rebuild per_node at that + * point in time because our resource groups aren't read in yet. + * The second time it's called is much later when we can rebuild it. + * + * allow_rebuild: 0 if rebuilds are not allowed + * 1 if rebuilds are allowed + */ +static void lookup_per_node(struct gfs2_sbd *sdp, int allow_rebuild) +{ + if (sdp->md.pinode) + return; + + gfs2_lookupi(sdp->master_dir, "per_node", 8, &sdp->md.pinode); + if (sdp->md.pinode) + return; + if (!allow_rebuild) { + log_err( _("The gfs2 system per_node directory " + "inode is missing, so we might not be \nable to " + "rebuild missing journals this run.\n")); + return; + } + + if (query( _("The gfs2 system per_node directory " + "inode is missing. Okay to rebuild it? (y/n) "))) { + int err; + + err = build_per_node(sdp); + if (err) { + log_crit(_("Error rebuilding per_node directory: %s\n"), + strerror(err)); + exit(-1); + } + } + gfs2_lookupi(sdp->master_dir, "per_node", 8, &sdp->md.pinode); + if (!sdp->md.pinode) { + log_err( _("Unable to rebuild per_node; aborting.\n")); + exit(-1); + } +} + +/** * init_system_inodes * * Returns: 0 on success, -1 on failure @@ -607,20 +653,9 @@ static int init_system_inodes(struct gfs2_sbd *sdp) } } - gfs2_lookupi(sdp->master_dir, "per_node", 8, &sdp->md.pinode); - if (!sdp->md.pinode) { - if (query( _("The gfs2 system per_node directory inode is " - "missing. Okay to rebuild it? (y/n) "))) { - err = build_per_node(sdp); - if (err) { - log_crit(_("Error rebuilding per_node directory: %s\n"), - strerror(err)); - exit(-1); - } - } - } - - /* FIXME fill in per_node structure */ + /* Try to lookup the per_node inode. If it was missing, it is now + safe to rebuild it. */ + lookup_per_node(sdp, 1); /******************************************************************* ******* Now, set boundary fields in the super block ************* *******************************************************************/ @@ -1219,6 +1254,11 @@ int initialize(struct gfs2_sbd *sbp, int force_check, int preen, sbp->sd_sb.sb_master_dir.no_addr); } + /* Look up the "per_node" inode. If there are journals missing, we + need to figure out what's missing from per_node. And we need all + our journals to be there before we can replay them. */ + lookup_per_node(sbp, 0); + /* verify various things */ if(replay_journals(sbp, preen, force_check, &clean_journals)) {

1 0

cluster: the annotated tag cluster-3.1.2 has been created
by Fabio M. Di Nitto 16 Jun '11

16 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=42… Commit: 42c2236dece1c5c6a94924f19f637fba62f2536d Parent: 0000000000000000000000000000000000000000 Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: 2011-06-16 13:40 +0000 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: 2011-06-16 13:40 +0000 annotated tag: cluster-3.1.2 has been created at 42c2236dece1c5c6a94924f19f637fba62f2536d (tag) tagging a548bd3333824341373a5f9a8ff9a456c5af59c8 (commit) replaces cluster-3.1.1 cluster-3.1.2 release David Teigland (1): dlm_controld: clear waiting plocks for closed files Fabio M. Di Nitto (6): cman-preconfig: allow cman to configure corosync multicast ttl cman init: increse the default timeout waiting for quorum cman init: add config option to disable _controld daemons startup config: add more totem options to relax ng schema for config validation cman preconfig: fix segfault if we cannot determine cluster name rgmanager: re-add slang scripts where they belong Lon Hohberger (5): qdiskd: Fix bad timer check cman: Don't increment on LEAVING->MEMBER transition Improve rgmanager's exclusive prioritization handling rgmanager: Pause during exit if we stopped services rgmanager: Fix reference count handling Masatake YAMATO (1): cman_tool: fix typo in man page Shane Bradley (1): cman: Clarify 'expected' option Vladislav Bogdanov (1): cman: do not add mcast address configuration to objdb when udpu is in use

1 0

cluster: RHEL6 - gfs2_edit: Fix savemeta compression for older zlibs
by Andrew Price 15 Jun '11

15 Jun '11

Gitweb: http://git.fedorahosted.org/git/cluster.git?p=cluster.git;a=commitdiff;h=d0… Commit: d02df4bcd1aee3ed287ee8f8c08c34c47db4cd1e Parent: 997c01bbc3c808f08879f334cd64d93e6d73fc0b Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Wed Jun 15 13:09:41 2011 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Wed Jun 15 14:17:08 2011 +0100 gfs2_edit: Fix savemeta compression for older zlibs gzdopen in zlib 1.2.3 parses the mode string in a different way to subsequent versions and the mode string we use causes the older gzwrite to fail with EBADF. This patch fixes the mode string so that the gzwrites succeed with the old and new zlibs. rhbz#702313 Reviewed-by: Steven Whitehouse <swhiteho(a)redhat.com> Signed-off-by: Andrew Price <anprice(a)redhat.com> --- gfs2/edit/savemeta.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/gfs2/edit/savemeta.c b/gfs2/edit/savemeta.c index cb7e243..160277c 100644 --- a/gfs2/edit/savemeta.c +++ b/gfs2/edit/savemeta.c @@ -209,7 +209,7 @@ static void warm_fuzzy_stuff(uint64_t wfsblock, int force) static struct metafd savemetaopen(char *out_fn, int gziplevel) { struct metafd mfd; - char gzmode[5] = "rwb9"; + char gzmode[3] = "w9"; char dft_fn[] = DFT_SAVE_FILE; if (!out_fn) { @@ -232,7 +232,7 @@ static struct metafd savemetaopen(char *out_fn, int gziplevel) mfd.gziplevel = gziplevel; if (gziplevel > 0) { - gzmode[3] = '0' + gziplevel; + gzmode[1] = '0' + gziplevel; mfd.gzfd = gzdopen(mfd.fd, gzmode); if (!mfd.gzfd) { fprintf(stderr, "gzdopen error: %s\n", strerror(errno));

1 0

gfs2-utils: master - gfs2_edit: Fix savemeta compression for older zlibs
by Andrew Price 15 Jun '11

15 Jun '11

Gitweb: http://git.fedorahosted.org/git/gfs2-utils.git?p=gfs2-utils.git;a=commitdif… Commit: 8331ddfd39725e0d206633488cf5582c252e9aa5 Parent: 68f8a96419c8a60f2598bc2b8ea2830fb206a2ca Author: Andrew Price <anprice(a)redhat.com> AuthorDate: Wed Jun 15 13:09:41 2011 +0100 Committer: Andrew Price <anprice(a)redhat.com> CommitterDate: Wed Jun 15 13:09:41 2011 +0100 gfs2_edit: Fix savemeta compression for older zlibs gzdopen in zlib 1.2.3 parses the mode string in a different way to subsequent versions and the mode string we use causes the older gzwrite to fail with EBADF. This patch fixes the mode string so that the gzwrites succeed with the old and new zlibs. Signed-off-by: Andrew Price <anprice(a)redhat.com> --- gfs2/edit/savemeta.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/gfs2/edit/savemeta.c b/gfs2/edit/savemeta.c index 1797568..0c567da 100644 --- a/gfs2/edit/savemeta.c +++ b/gfs2/edit/savemeta.c @@ -211,7 +211,7 @@ static void warm_fuzzy_stuff(uint64_t wfsblock, int force) static struct metafd savemetaopen(char *out_fn, int gziplevel) { struct metafd mfd; - char gzmode[5] = "rwb9"; + char gzmode[3] = "w9"; char dft_fn[] = DFT_SAVE_FILE; if (!out_fn) { @@ -234,7 +234,7 @@ static struct metafd savemetaopen(char *out_fn, int gziplevel) mfd.gziplevel = gziplevel; if (gziplevel > 0) { - gzmode[3] = '0' + gziplevel; + gzmode[1] = '0' + gziplevel; mfd.gzfd = gzdopen(mfd.fd, gzmode); if (!mfd.gzfd) { fprintf(stderr, "gzdopen error: %s\n", strerror(errno));

1 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

cluster-commits June 2011