Re: [Qemu-devel] [PATCH v2 2/5] block: Use error code EMEDIUMTYPE for wrong format in some block drivers

[Stefan Weil <sw@weilnetz.de>]

Am 18.01.2013 09:53, schrieb Markus Armbruster:
> Stefan Weil <sw@weilnetz.de> writes:
>> This improves error reports for bochs, cow, qcow, qcow2, qed and vmdk
>> when a file with the wrong format is selected.
>>
>> Signed-off-by: Stefan Weil <sw@weilnetz.de>
>> ---
>>  block/bochs.c |    2 +-
>>  block/cow.c   |    2 +-
>>  block/qcow.c  |    2 +-
>>  block/qcow2.c |    2 +-
>>  block/qed.c   |    2 +-
>>  block/vmdk.c  |    4 ++--
>>  6 files changed, 7 insertions(+), 7 deletions(-)
>>
>> diff --git a/block/bochs.c b/block/bochs.c
>> index 1b1d9cd..3737583 100644
>> --- a/block/bochs.c
>> +++ b/block/bochs.c
>> @@ -126,7 +126,7 @@ static int bochs_open(BlockDriverState *bs, int flags)
>>          strcmp(bochs.subtype, GROWING_TYPE) ||
>>  	((le32_to_cpu(bochs.version) != HEADER_VERSION) &&
>>  	(le32_to_cpu(bochs.version) != HEADER_V1))) {
>> -        goto fail;
>> +        return -EMEDIUMTYPE;
>>      }
>>  
>>      if (le32_to_cpu(bochs.version) == HEADER_V1) {
> You make the function return either 0, -1 or -EMEDIUMTYPE.  Please make
> it return either 0 or a negative errno code, like this (untested):

Hi Markus,

returning 0, -1 is like before, only returning -EMEDIUMTYPE is new.

You are right, a return value of -1 should be replaced by a negative
error value. I fixed this for block/vdi.c in a separate patch as
suggested by Kevin, see http://patchwork.ozlabs.org/patch/213375/.

The same kind of improvement should be done for other block
drivers which currently use -1, but that can be done after my
patch series was applied.

The primary purpose of my patch series was fixing open bugreports.
For vdi I did more because I feel responsible for that part of the
code.

Regards,
StefanW.