On Mon, Mar 4, 2013 at 10:27 AM, Corey Bryant <cor...@linux.vnet.ibm.com> wrote:
>
>
> On 03/02/2013 01:58 AM, Doug Goldstein wrote:
>>
>> Handle errors and cleanup from the error in a unified place for
>> parse_acl_file().
>>
>> Signed-off-by: Doug Goldstein <car...@cardoe.com>
>> CC: Anthony Liguori <aligu...@us.ibm.com>
>> CC: Richa Marwaha <rmar...@linux.vnet.ibm.com>
>> CC: Corey Bryant <cor...@linux.vnet.ibm.com>
>> TO: qemu-devel@nongnu.org
>> ---
>> qemu-bridge-helper.c | 20 +++++++++++---------
>> 1 file changed, 11 insertions(+), 9 deletions(-)
>>
>> diff --git a/qemu-bridge-helper.c b/qemu-bridge-helper.c
>> index 287bfd5..ee67740 100644
>> --- a/qemu-bridge-helper.c
>> +++ b/qemu-bridge-helper.c
>> @@ -74,11 +74,12 @@ static int parse_acl_file(const char *filename,
>> ACLList *acl_list)
>> {
>> FILE *f;
>> char line[4096];
>> + int ret = -EINVAL;
>> ACLRule *acl_rule;
>>
>> f = fopen(filename, "r");
>> if (f == NULL) {
>> - return -1;
>> + return -errno;
>> }
>>
>> while (fgets(line, sizeof(line), f) != NULL) {
>> @@ -102,9 +103,8 @@ static int parse_acl_file(const char *filename,
>> ACLList *acl_list)
>>
>> if (arg == NULL) {
>> fprintf(stderr, "Invalid config line:\n %s\n", line);
>> - fclose(f);
>> - errno = EINVAL;
>> - return -1;
>> + ret = -EINVAL;
>> + goto failure;
>
>
> I would stick with setting errno here rather than ret..
>
>
>> }
>>
>> *arg = 0;
>> @@ -142,15 +142,17 @@ static int parse_acl_file(const char *filename,
>> ACLList *acl_list)
>> parse_acl_file(arg, acl_list);
>> } else {
>> fprintf(stderr, "Unknown command `%s'\n", cmd);
>> - fclose(f);
>> - errno = EINVAL;
>> - return -1;
>> + ret = -EINVAL;
>> + goto failure;
>
>
> And do the same here..
>
>
>> }
>> }
>>
>> + ret = 0;
>> +
>> +failure:
>> fclose(f);
>>
>> - return 0;
>> + return ret;
>> }
>>
>> static bool has_vnet_hdr(int fd)
>> @@ -272,7 +274,7 @@ int main(int argc, char **argv)
>>
>> /* parse default acl file */
>> QSIMPLEQ_INIT(&acl_list);
>> - if (parse_acl_file(DEFAULT_ACL_FILE, &acl_list) == -1) {
>> + if (parse_acl_file(DEFAULT_ACL_FILE, &acl_list) < 0) {
>> fprintf(stderr, "failed to parse default acl file `%s'\n",
>> DEFAULT_ACL_FILE);
>
>
> .. and then you can append strerror(errno) to this message, which I admit
> should have been here before you touched this code. This will keep this
> error path consistent with many of the others in this file.
Would you consider the return value then being passed on to
strerror()? Seems like it'd be a little bit safer from the stand point
of someone coming through and adding something new the the cleanup
case or any other cases which calls a glibc function which then blows
away errno.
--
Doug Goldstein
