[for-2.9] block/archipelago: Make it compile
diff mbox

Message ID 20170308181833.32110-1-mreitz@redhat.com
State New
Headers show

Commit Message

Max Reitz March 8, 2017, 6:18 p.m. UTC
In order to use error_setg() and similar functions, we need to include
qapi/error.h.

Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/archipelago.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Eric Blake March 8, 2017, 6:25 p.m. UTC | #1
On 03/08/2017 12:18 PM, Max Reitz wrote:
> In order to use error_setg() and similar functions, we need to include
> qapi/error.h.
> 
> Signed-off-by: Max Reitz <mreitz@redhat.com>
> ---
>  block/archipelago.c | 1 +
>  1 file changed, 1 insertion(+)

Reviewed-by: Eric Blake <eblake@redhat.com>

Does this mean our automated buildbots aren't building archipelago by
default?

> 
> diff --git a/block/archipelago.c b/block/archipelago.c
> index 2449cfc702..01c4ff2f7d 100644
> --- a/block/archipelago.c
> +++ b/block/archipelago.c
> @@ -55,6 +55,7 @@
>  #include "block/block_int.h"
>  #include "qemu/error-report.h"
>  #include "qemu/thread.h"
> +#include "qapi/error.h"
>  #include "qapi/qmp/qint.h"
>  #include "qapi/qmp/qstring.h"
>  #include "qapi/qmp/qjson.h"
>
Max Reitz March 8, 2017, 6:27 p.m. UTC | #2
On 08.03.2017 19:25, Eric Blake wrote:
> On 03/08/2017 12:18 PM, Max Reitz wrote:
>> In order to use error_setg() and similar functions, we need to include
>> qapi/error.h.
>>
>> Signed-off-by: Max Reitz <mreitz@redhat.com>
>> ---
>>  block/archipelago.c | 1 +
>>  1 file changed, 1 insertion(+)
> 
> Reviewed-by: Eric Blake <eblake@redhat.com>
> 
> Does this mean our automated buildbots aren't building archipelago by
> default?

Well, at least patchew didn't catch that I definitely broke archipelago
in my bdrv_truncate() errp series...

[CC-ing Fam]

Max
Eric Blake March 8, 2017, 6:28 p.m. UTC | #3
On 03/08/2017 12:25 PM, Eric Blake wrote:
> On 03/08/2017 12:18 PM, Max Reitz wrote:
>> In order to use error_setg() and similar functions, we need to include
>> qapi/error.h.
>>
>> Signed-off-by: Max Reitz <mreitz@redhat.com>
>> ---
>>  block/archipelago.c | 1 +
>>  1 file changed, 1 insertion(+)
> 
> Reviewed-by: Eric Blake <eblake@redhat.com>
> 
> Does this mean our automated buildbots aren't building archipelago by
> default?
> 

Oh, I meant to add:

Should probably mention that it was commit da34e65 that introduced the
problem (if I'm right?).  Does that mean we have not been compiling
archipelago.c since Mar 2016?
Max Reitz March 8, 2017, 6:32 p.m. UTC | #4
On 08.03.2017 19:28, Eric Blake wrote:
> On 03/08/2017 12:25 PM, Eric Blake wrote:
>> On 03/08/2017 12:18 PM, Max Reitz wrote:
>>> In order to use error_setg() and similar functions, we need to include
>>> qapi/error.h.
>>>
>>> Signed-off-by: Max Reitz <mreitz@redhat.com>
>>> ---
>>>  block/archipelago.c | 1 +
>>>  1 file changed, 1 insertion(+)
>>
>> Reviewed-by: Eric Blake <eblake@redhat.com>
>>
>> Does this mean our automated buildbots aren't building archipelago by
>> default?
>>
> 
> Oh, I meant to add:
> 
> Should probably mention that it was commit da34e65 that introduced the
> problem (if I'm right?).  Does that mean we have not been compiling
> archipelago.c since Mar 2016?

At least here on my machine it does indeed compile before that commit
and fails afterwards. So, yes, maybe we actually have not. :-)

Max
Philippe Mathieu-Daudé March 8, 2017, 6:38 p.m. UTC | #5
On 03/08/2017 03:25 PM, Eric Blake wrote:
> On 03/08/2017 12:18 PM, Max Reitz wrote:
>> In order to use error_setg() and similar functions, we need to include
>> qapi/error.h.
>>
>> Signed-off-by: Max Reitz <mreitz@redhat.com>
>> ---
>>  block/archipelago.c | 1 +
>>  1 file changed, 1 insertion(+)
>
> Reviewed-by: Eric Blake <eblake@redhat.com>

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

>
> Does this mean our automated buildbots aren't building archipelago by
> default?
>
>>
>> diff --git a/block/archipelago.c b/block/archipelago.c
>> index 2449cfc702..01c4ff2f7d 100644
>> --- a/block/archipelago.c
>> +++ b/block/archipelago.c
>> @@ -55,6 +55,7 @@
>>  #include "block/block_int.h"
>>  #include "qemu/error-report.h"
>>  #include "qemu/thread.h"
>> +#include "qapi/error.h"
>>  #include "qapi/qmp/qint.h"
>>  #include "qapi/qmp/qstring.h"
>>  #include "qapi/qmp/qjson.h"
>>
>
Kevin Wolf March 8, 2017, 6:46 p.m. UTC | #6
Am 08.03.2017 um 19:32 hat Max Reitz geschrieben:
> On 08.03.2017 19:28, Eric Blake wrote:
> > On 03/08/2017 12:25 PM, Eric Blake wrote:
> >> On 03/08/2017 12:18 PM, Max Reitz wrote:
> >>> In order to use error_setg() and similar functions, we need to include
> >>> qapi/error.h.
> >>>
> >>> Signed-off-by: Max Reitz <mreitz@redhat.com>
> >>> ---
> >>>  block/archipelago.c | 1 +
> >>>  1 file changed, 1 insertion(+)
> >>
> >> Reviewed-by: Eric Blake <eblake@redhat.com>
> >>
> >> Does this mean our automated buildbots aren't building archipelago by
> >> default?
> >>
> > 
> > Oh, I meant to add:
> > 
> > Should probably mention that it was commit da34e65 that introduced the
> > problem (if I'm right?).  Does that mean we have not been compiling
> > archipelago.c since Mar 2016?
> 
> At least here on my machine it does indeed compile before that commit
> and fails afterwards. So, yes, maybe we actually have not. :-)

In that case, maybe we should rather remove the driver because very
obviously nobody is interested in it, not even the maintainer. Keeping
things around that nobody tests (and consequently, nobody uses) is just
a useless maintenance burden.

Kevin
Eric Blake March 8, 2017, 7:33 p.m. UTC | #7
On 03/08/2017 12:46 PM, Kevin Wolf wrote:
>>> Oh, I meant to add:
>>>
>>> Should probably mention that it was commit da34e65 that introduced the
>>> problem (if I'm right?).  Does that mean we have not been compiling
>>> archipelago.c since Mar 2016?
>>
>> At least here on my machine it does indeed compile before that commit
>> and fails afterwards. So, yes, maybe we actually have not. :-)
> 
> In that case, maybe we should rather remove the driver because very
> obviously nobody is interested in it, not even the maintainer. Keeping
> things around that nobody tests (and consequently, nobody uses) is just
> a useless maintenance burden.

We've released 2.6, 2.7, and 2.8 with no complaints about failure to
compile. Sounds like a good candidate for removal in 2.9.

Patch
diff mbox

diff --git a/block/archipelago.c b/block/archipelago.c
index 2449cfc702..01c4ff2f7d 100644
--- a/block/archipelago.c
+++ b/block/archipelago.c
@@ -55,6 +55,7 @@ 
 #include "block/block_int.h"
 #include "qemu/error-report.h"
 #include "qemu/thread.h"
+#include "qapi/error.h"
 #include "qapi/qmp/qint.h"
 #include "qapi/qmp/qstring.h"
 #include "qapi/qmp/qjson.h"