All of lore.kernel.org
 help / color / mirror / Atom feed
* Re: scsi: lpfc_scsi: Remove unused variables
       [not found] <4FD0434B.7090405@gentoo.org>
@ 2012-06-07  6:12 ` Jonathan Nieder
  2012-06-07  6:21   ` Richard Yao
  2012-06-07  6:37   ` Greg KH
  0 siblings, 2 replies; 5+ messages in thread
From: Jonathan Nieder @ 2012-06-07  6:12 UTC (permalink / raw)
  To: Richard Yao; +Cc: stable, linux-scsi, James Smart

Hi Richard,

Richard Yao wrote:

> This is my first time patch submission. The commit message should
> explain the issue.
>
> commit ba8c7a7a629b736fdc1abceaa3ec49f8820aec08
> Author: Richard Yao <ryao@cs.stonybrook.edu>
> Date:   Thu Jun 7 01:44:31 2012 -0400
>
>     scsi: lpfc_scsi: Remove unused variables
>
>     Kernels built with CONFIG_SCSI_LPFC fail to compile due to the use of
>     -Werror=unused-variable. We solve that by removing some unused
>     variables.

Wouldn't a saner fix be to stop using -Werror=unused-variable, or am I
missing something subtle?

(Since v3.4-rc1~6^2~48 the makefile seems to use plain -Werror, which
seems even worse, since it means any random compiler bug that triggers
a new warning will break the build.)

Thanks and hope that helps,
Jonathan

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: scsi: lpfc_scsi: Remove unused variables
  2012-06-07  6:12 ` scsi: lpfc_scsi: Remove unused variables Jonathan Nieder
@ 2012-06-07  6:21   ` Richard Yao
  2012-06-07  6:42     ` Jonathan Nieder
  2012-06-07  6:37   ` Greg KH
  1 sibling, 1 reply; 5+ messages in thread
From: Richard Yao @ 2012-06-07  6:21 UTC (permalink / raw)
  To: Jonathan Nieder; +Cc: stable, linux-scsi, James Smart

On 06/07/12 02:12, Jonathan Nieder wrote:
> Hi Richard,
>
> Richard Yao wrote:
>
>> This is my first time patch submission. The commit message should
>> explain the issue.
>>
>> commit ba8c7a7a629b736fdc1abceaa3ec49f8820aec08
>> Author: Richard Yao <ryao@cs.stonybrook.edu>
>> Date:   Thu Jun 7 01:44:31 2012 -0400
>>
>>     scsi: lpfc_scsi: Remove unused variables
>>
>>     Kernels built with CONFIG_SCSI_LPFC fail to compile due to the use of
>>     -Werror=unused-variable. We solve that by removing some unused
>>     variables.
> Wouldn't a saner fix be to stop using -Werror=unused-variable, or am I
> missing something subtle?
>
> (Since v3.4-rc1~6^2~48 the makefile seems to use plain -Werror, which
> seems even worse, since it means any random compiler bug that triggers
> a new warning will break the build.)
>
> Thanks and hope that helps,
> Jonathan

Using -Werror enables us to catch potential runtime issues before they
happen. This one is not a runtime issue, but the compiler is right in
stating that these variables serve no purpose. My opinion is that
occasional noise due to -Werror won't be a problem. Also, decisions on
-Werror probably should be made by either the subsystem maintainer or
the branch maintainer.

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: scsi: lpfc_scsi: Remove unused variables
  2012-06-07  6:12 ` scsi: lpfc_scsi: Remove unused variables Jonathan Nieder
  2012-06-07  6:21   ` Richard Yao
@ 2012-06-07  6:37   ` Greg KH
  1 sibling, 0 replies; 5+ messages in thread
From: Greg KH @ 2012-06-07  6:37 UTC (permalink / raw)
  To: Jonathan Nieder; +Cc: Richard Yao, stable, linux-scsi, James Smart

On Thu, Jun 07, 2012 at 01:12:56AM -0500, Jonathan Nieder wrote:
> Hi Richard,
> 
> Richard Yao wrote:
> 
> > This is my first time patch submission. The commit message should
> > explain the issue.
> >
> > commit ba8c7a7a629b736fdc1abceaa3ec49f8820aec08
> > Author: Richard Yao <ryao@cs.stonybrook.edu>
> > Date:   Thu Jun 7 01:44:31 2012 -0400
> >
> >     scsi: lpfc_scsi: Remove unused variables
> >
> >     Kernels built with CONFIG_SCSI_LPFC fail to compile due to the use of
> >     -Werror=unused-variable. We solve that by removing some unused
> >     variables.
> 
> Wouldn't a saner fix be to stop using -Werror=unused-variable, or am I
> missing something subtle?

Yes, that is correct, older versions of gcc, and sometimes newer ones,
get this logic wrong and incorrectly flag variables that really are used
due to bugs in their detection logic.  Don't require a developer to have
a gcc version that is different from what the rest of the kernel
requires, as that is not allowed.

thanks,

greg k-h

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: scsi: lpfc_scsi: Remove unused variables
  2012-06-07  6:21   ` Richard Yao
@ 2012-06-07  6:42     ` Jonathan Nieder
  2012-06-11 14:02       ` James Smart
  0 siblings, 1 reply; 5+ messages in thread
From: Jonathan Nieder @ 2012-06-07  6:42 UTC (permalink / raw)
  To: Richard Yao; +Cc: stable, linux-scsi, James Smart

Richard Yao wrote:

> Using -Werror enables us to catch potential runtime issues before they
> happen.

You have more trust in GCC and other compilers than I do.

If you'd like some entertainment, then for example see:

 http://gcc.gnu.org/PR52798
 http://gcc.gnu.org/PR25509

Bugs and misdesigned features do not have to be involved for this to
be a problem --- all it takes is a gcc upgrade that adds a new
warning, and suddenly you've turned your existing potential code
cleanliness improvements into failed builds and lots of wasted time
for innocent sysadmins.  -Werror is great for development, but using
it in production turns out to be pretty much always a terrible idea.

Thanks for the chance to vent about this. :)  It's bad, bad, bad.

Ciao,
Jonathan

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: scsi: lpfc_scsi: Remove unused variables
  2012-06-07  6:42     ` Jonathan Nieder
@ 2012-06-11 14:02       ` James Smart
  0 siblings, 0 replies; 5+ messages in thread
From: James Smart @ 2012-06-11 14:02 UTC (permalink / raw)
  To: Jonathan Nieder
  Cc: Richard Yao, stable@vger.kernel.org, linux-scsi@vger.kernel.org

Richard,

fyi - this error in lpfc was corrected in our 8.3.31 patch set - posted 
back in May:
http://www.spinics.net/lists/linux-scsi/msg59216.html

-- james


On 6/7/2012 2:42 AM, Jonathan Nieder wrote:
> Richard Yao wrote:
>
>> Using -Werror enables us to catch potential runtime issues before they
>> happen.
> You have more trust in GCC and other compilers than I do.
>
> If you'd like some entertainment, then for example see:
>
>   http://gcc.gnu.org/PR52798
>   http://gcc.gnu.org/PR25509
>
> Bugs and misdesigned features do not have to be involved for this to
> be a problem --- all it takes is a gcc upgrade that adds a new
> warning, and suddenly you've turned your existing potential code
> cleanliness improvements into failed builds and lots of wasted time
> for innocent sysadmins.  -Werror is great for development, but using
> it in production turns out to be pretty much always a terrible idea.
>
> Thanks for the chance to vent about this. :)  It's bad, bad, bad.
>
> Ciao,
> Jonathan
>

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2012-06-11 14:02 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <4FD0434B.7090405@gentoo.org>
2012-06-07  6:12 ` scsi: lpfc_scsi: Remove unused variables Jonathan Nieder
2012-06-07  6:21   ` Richard Yao
2012-06-07  6:42     ` Jonathan Nieder
2012-06-11 14:02       ` James Smart
2012-06-07  6:37   ` Greg KH

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.