linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* feature-request
@ 2008-02-23 13:39 Pierre Habouzit
  2008-02-25  2:48 ` feature-request Pavel Roskin
  0 siblings, 1 reply; 4+ messages in thread
From: Pierre Habouzit @ 2008-02-23 13:39 UTC (permalink / raw)
  To: linux-sparse

[-- Attachment #1: Type: text/plain, Size: 539 bytes --]

  While I'm at it, there is this feature I'd like to see in sparse: I'd
love to be able to ask it to ignore errors that are located in some
specific paths (like /usr/include e.g.). For now I'm doing that through
a custom script, but it'd be simpler for me if it does it natively.  The
reason is that I don't want to patch third party libraries headers.


-- 
·O·  Pierre Habouzit
··O                                                madcoder@debian.org
OOO                                                http://www.madism.org

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: feature-request
  2008-02-23 13:39 feature-request Pierre Habouzit
@ 2008-02-25  2:48 ` Pavel Roskin
  2008-02-25  8:54   ` feature-request Pierre Habouzit
  0 siblings, 1 reply; 4+ messages in thread
From: Pavel Roskin @ 2008-02-25  2:48 UTC (permalink / raw)
  To: Pierre Habouzit; +Cc: linux-sparse


On Sat, 2008-02-23 at 14:39 +0100, Pierre Habouzit wrote:
> While I'm at it, there is this feature I'd like to see in sparse: I'd
> love to be able to ask it to ignore errors that are located in some
> specific paths (like /usr/include e.g.). For now I'm doing that through
> a custom script, but it'd be simpler for me if it does it natively.  The
> reason is that I don't want to patch third party libraries headers.

We probably don't want to ignore _errors_, as they can indicate that the
parser doesn't understand the code correctly.  Ignoring warnings would
be a good idea.

Perhaps a simpler approach would be to turn off warnings in any files
included using angle brackets.

gcc doesn't report warnings in system files by default, but it can be
enabled with -Wsystem-headers.  Since sparse is primarily for the
kernel, I think the default should be to check the headers (as they are
part of the kernel), but sparse could support -Wno-system-headers.

Alternatively, sparse could have a userspace mode that would disable
warnings in system headers by default (unless -Wsystem-headers is used).
The kernel mode could be made stricter for the kernel without affecting
the userspace.

-- 
Regards,
Pavel Roskin

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: feature-request
  2008-02-25  2:48 ` feature-request Pavel Roskin
@ 2008-02-25  8:54   ` Pierre Habouzit
  2008-02-25 14:22     ` feature-request Pavel Roskin
  0 siblings, 1 reply; 4+ messages in thread
From: Pierre Habouzit @ 2008-02-25  8:54 UTC (permalink / raw)
  To: Pavel Roskin; +Cc: linux-sparse

[-- Attachment #1: Type: text/plain, Size: 1895 bytes --]

On Mon, Feb 25, 2008 at 02:48:45AM +0000, Pavel Roskin wrote:
> 
> On Sat, 2008-02-23 at 14:39 +0100, Pierre Habouzit wrote:
> > While I'm at it, there is this feature I'd like to see in sparse: I'd
> > love to be able to ask it to ignore errors that are located in some
> > specific paths (like /usr/include e.g.). For now I'm doing that through
> > a custom script, but it'd be simpler for me if it does it natively.  The
> > reason is that I don't want to patch third party libraries headers.
> 
> We probably don't want to ignore _errors_, as they can indicate that the
> parser doesn't understand the code correctly.  Ignoring warnings would
> be a good idea.

  Of course it's what I meant.

> Perhaps a simpler approach would be to turn off warnings in any files
> included using angle brackets.

  Nope, that's not good, because I use angle brackets to #include files
from my projects when I use internally in-tree headers that will in the
end be public.

  I'd rather like to ask to ignore warnings for example, for file under
/usr/include and /usr/lib/gcc/

> gcc doesn't report warnings in system files by default, but it can be
> enabled with -Wsystem-headers.  Since sparse is primarily for the
> kernel, I think the default should be to check the headers (as they are
> part of the kernel), but sparse could support -Wno-system-headers.

  yeah, that'd be really great.

> Alternatively, sparse could have a userspace mode that would disable
> warnings in system headers by default (unless -Wsystem-headers is used).
> The kernel mode could be made stricter for the kernel without affecting
> the userspace.

  I don't really mind this way or the other, really :)

-- 
·O·  Pierre Habouzit
··O                                                madcoder@debian.org
OOO                                                http://www.madism.org

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: feature-request
  2008-02-25  8:54   ` feature-request Pierre Habouzit
@ 2008-02-25 14:22     ` Pavel Roskin
  0 siblings, 0 replies; 4+ messages in thread
From: Pavel Roskin @ 2008-02-25 14:22 UTC (permalink / raw)
  To: Pierre Habouzit; +Cc: linux-sparse

On Mon, 2008-02-25 at 09:54 +0100, Pierre Habouzit wrote:

> > Perhaps a simpler approach would be to turn off warnings in any files
> > included using angle brackets.
> 
>   Nope, that's not good, because I use angle brackets to #include files
> from my projects when I use internally in-tree headers that will in the
> end be public.
> 
>   I'd rather like to ask to ignore warnings for example, for file under
> /usr/include and /usr/lib/gcc/

You are right.  I think sparse should try to follow gcc in this regard.
There are three hardcoded directories in pre-process.c
(/usr/include, /usr/local/include and the gcc include directory).  Maybe
they should be considered system include directories.

-- 
Regards,
Pavel Roskin

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2008-02-25 14:22 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2008-02-23 13:39 feature-request Pierre Habouzit
2008-02-25  2:48 ` feature-request Pavel Roskin
2008-02-25  8:54   ` feature-request Pierre Habouzit
2008-02-25 14:22     ` feature-request Pavel Roskin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).