On 17/11/11 09:31, Richard W.M. Jones wrote:
On Wed, Nov 16, 2011 at 05:54:24PM +0000, Matthew Booth wrote:
>
> Return a hash containing metadata about a specific Linux MD device, based on the
> output of 'mdadm -DY'.
> ---
> daemon/md.c | 78 ++++++++++++++++++++++++++++++++++++++++
> generator/generator_actions.ml | 31 ++++++++++++++++
> regressions/test-mdadm.sh | 60 ++++++++++++++++++++++++++++++-
> src/MAX_PROC_NR | 2 +-
> 4 files changed, 169 insertions(+), 2 deletions(-)
>
> diff --git a/daemon/md.c b/daemon/md.c
> index 257bd0f..c1f9d2a 100644
> --- a/daemon/md.c
> +++ b/daemon/md.c
> @@ -23,6 +23,7 @@
> #include<string.h>
> #include<inttypes.h>
> #include<glob.h>
> +#include<ctype.h>
>
> #include "daemon.h"
> #include "actions.h"
> @@ -230,3 +231,80 @@ error:
> if (r != NULL) free_strings (r);
> return NULL;
> }
> +
> +char **
> +do_mdadm_detail(const char *md)
> +{
> + int r;
> + char *out, *err;
> +
> + char **ret = NULL;
> + int size = 0, alloc = 0;
> +
> + const char *mdadm[] = { "mdadm", "-DY", md, NULL };
> + r = commandv(&out,&err, mdadm);
Any reason not to use plain ol' "command" here? It takes a
variable number of args.
It's slightly more efficient under the hood (no need for malloc) and not
especially untidy. I have no strong opinion.
> + if (r == -1) {
> + reply_with_error("%s", err);
> + free(err);
> + free(out);
> + return NULL;
> + }
Somewhere about here, free(err) needs to be called.
I was using some of your code from ext2.c as a template here. I hadn't
actually bothered to check, but I assumed that as err wasn't freed, it
was never assigned. Checking the code, it seems it is always allocated,
which means there's the same leak in do_tune2fs_l.
I'll update it here.
> + char *p = out;
> +
> + /* Parse the output of mdadm -DY:
> + * MD_LEVEL=raid1
> + * MD_DEVICES=2
> + * MD_METADATA=1.0
> + * MD_UUID=cfa81b59:b6cfbd53:3f02085b:58f4a2e1
> + * MD_NAME=localhost.localdomain:0
> + */
> + while (*p) {
> + /* Turn the newline (if any) into a null */
> + char *nl = strchrnul(p, '\n');
> + if (*nl == '\n') {
> + *nl = '\0';
> + nl++;
> + }
There is a function 'split_lines' defined in daemon.h which
essentially does this (not skipping blank lines though, but
I don't think those are really possible in mdadm output).
Looks good. I'll use it. You may want to consider using it in do_tune2fs_l.
> + /* Skip blank lines (shouldn't happen) */
> + if (!*p) continue;
> +
> + /* Split the line in 2 at the equals sign */
> + char *eq = strchr(p, '=');
> + if (eq) {
> + *eq = '\0'; eq++;
> +
> + /* Remove the MD_ prefix from the key and translate the remainder to lower
> + * case */
> + if (STRPREFIX(p, "MD_")) {
> + p += 3;
> + for (char *i = p; *i != '\0'; i++) {
> + *i = tolower(*i);
Probably c_tolower (from "c-ctype.h") is better to use here, since
you're really dealing with 7-bit ASCII in octets, not characters.
Ok.
> + }
> + }
> +
> + /* Add the key/value pair to the output */
> + if (add_string(&ret,&size,&alloc, p) == -1) {
> + free(out);
> + return NULL;
> + }
> + if (add_string(&ret,&size,&alloc, eq) == -1) {
> + free(out);
> + return NULL;
> + }
> + } else {
> + /* Ignore lines with no equals sign (shouldn't happen). Log to stderr so
> + * it will show up in LIBGUESTFS_DEBUG. */
> + fprintf(stderr, "Unexpected output from mdadm -DY: %s", p);
It's more consistent if the error message is something like:
"mdadm-delete: unexpected output ignored: %s"
Ok.
> + }
> +
> + p = nl;
> + }
> +
> + free(out);
> +
> + if (add_string(&ret,&size,&alloc, NULL) == -1) return NULL;
> +
> + return ret;
> +}
> diff --git a/generator/generator_actions.ml b/generator/generator_actions.ml
> index a4658a0..9bf9def 100644
> --- a/generator/generator_actions.ml
> +++ b/generator/generator_actions.ml
> @@ -6496,6 +6496,37 @@ If not set, this defaults to C<raid1>.
> "\
> List all Linux md devices.");
>
> + ("mdadm_detail", (RHashtable "info", [Device "md"],
[]), 301, [Optional "mdadm"],
> + [],
> + "obtain metadata for an MD device",
> + "\
> +This command exposes the output of 'mdadm -DY<md>'. The resulting hash
will
> +contain at least:
I wouldn't make any specific claims about what mdadm can return. I
would say something like:
"The following fields are usually present in the returned hash.
Other fields may also be present."
Ok.
> +=over
> +
> +=item C<level>
> +
> +The raid level of the MD device.
> +
> +=item C<devices>
> +
> +The number of underlying devices in the MD device.
> +
> +=item C<metadata>
> +
> +The metadata version used.
> +
> +=item C<uuid>
> +
> +The UUID of the MD device.
> +
> +=item C<name>
> +
> +The name of the MD device.
> +
> +=back");
> +
> ]
>
> let all_functions = non_daemon_functions @ daemon_functions
> diff --git a/regressions/test-mdadm.sh b/regressions/test-mdadm.sh
> index 3ad4f22..8119561 100755
> --- a/regressions/test-mdadm.sh
> +++ b/regressions/test-mdadm.sh
> @@ -92,4 +92,62 @@ write /r5t3/baz "testing"
>
> EOF
>
> -rm -f md-test1.img md-test2.img md-test3.img md-test4.img
> +eval `../fish/guestfish --listen`
> +../fish/guestfish --remote add-ro md-test1.img
> +../fish/guestfish --remote add-ro md-test2.img
> +../fish/guestfish --remote add-ro md-test3.img
> +../fish/guestfish --remote add-ro md-test4.img
> +../fish/guestfish --remote run
> +
> +for md in `../fish/guestfish --remote list-md-devices`; do
> + ../fish/guestfish --remote mdadm-detail "${md}"> mdadm-detail.out
> +
> + sed 's/:\s*/=/' mdadm-detail.out> mdadm-detail.out.sh
> + . mdadm-detail.out.sh
> + rm -f mdadm-detail.out.sh
> +
> + error=0
> + case "$name" in
> + *:r1t1)
> + [ "$level" == "raid1" ] || error=1
> + [ "$devices" == "2" ] || error=1
> + ;;
> +
> + *:r1t2)
> + [ "$level" == "raid1" ] || error=1
> + [ "$devices" == "2" ] || error=1
> + ;;
> +
> + *:r5t1)
> + [ "$level" == "raid5" ] || error=1
> + [ "$devices" == "4" ] || error=1
> + ;;
> +
> + *:r5t2)
> + [ "$level" == "raid5" ] || error=1
> + [ "$devices" == "3" ] || error=1
> + ;;
> +
> + *:r5t3)
> + [ "$level" == "raid5" ] || error=1
> + [ "$devices" == "2" ] || error=1
> + ;;
> +
> + *)
> + error=1
> + esac
> +
> + [[ "$uuid" =~ ([0-9a-f]{8}:){3}[0-9a-f]{8} ]] || error=1
> + [ ! -z "$metadata" ] || error=1
> +
> + if [ "$error" == "1" ]; then
> + echo "$0: Unexpected output from mdadm-detail for device $md"
> + cat mdadm-detail.out
> + ../fish/guestfish --remote exit
> + exit 1
> + fi
> +done
> +
> +../fish/guestfish --remote exit
> +
> +rm -f mdadm-detail.out md-test1.img md-test2.img md-test3.img md-test4.img
Looks good. Did the test run OK?
Yes. The test ran fine.
Updated patch to follow. Thanks,
Matt
--
Matthew Booth, RHCA, RHCSS
Red Hat Engineering, Virtualisation Team
GPG ID: D33C3490
GPG FPR: 3733 612D 2D05 5458 8A8A 1600 3441 EA19 D33C 3490