[Qemu-devel] [PATCH][RFC] Fix gcc 4.x warnings (-Wpointer-sign)

[Qemu-devel] [PATCH][RFC] Fix gcc 4.x warnings (-Wpointer-sign)

[Blue Swirl]

[-- Attachment #1: Type: text/plain, Size: 121 bytes --]

Hi,

GCC 4 enables -Wpointer-sign if -Wall is enabled. This trivial patch
suppresses the warnings. Comments? Objections?

[-- Attachment #2: fix_signed_warnings.diff --]
[-- Type: plain/text, Size: 24496 bytes --]

Re: [Qemu-devel] [PATCH][RFC] Fix gcc 4.x warnings (-Wpointer-sign)

[M. Warner Losh]

In message: <f43fc5580809121256i1d15c407vbebbfc8cd081cb43@mail.gmail.com>
            "Blue Swirl" <blauwirbel@gmail.com> writes:
: GCC 4 enables -Wpointer-sign if -Wall is enabled. This trivial patch
: suppresses the warnings. Comments? Objections?

Don't you mean 'fix' the warnings, since this patch looks like it
tries to cast things to be correct...

Warner

Re: [Qemu-devel] [PATCH][RFC] Fix gcc 4.x warnings (-Wpointer-sign)

[Blue Swirl]

On 9/12/08, M. Warner Losh <imp@bsdimp.com> wrote:
> In message: <f43fc5580809121256i1d15c407vbebbfc8cd081cb43@mail.gmail.com>
>             "Blue Swirl" <blauwirbel@gmail.com> writes:
>  : GCC 4 enables -Wpointer-sign if -Wall is enabled. This trivial patch
>
> : suppresses the warnings. Comments? Objections?
>
>
> Don't you mean 'fix' the warnings, since this patch looks like it
>  tries to cast things to be correct...

Well, some parts of the patch actually fix the types so that they are
more consistent, the casts just silence the warnings. A real fix would
be to change the types but then it would be incorrect to change the
signed types just to match the unsigned loadvm/savevm API.

Now that I think of it, the API could be extended with signed
versions. That would be best in terms of type safety. Maybe I'll try
that approach instead.

Re: [Qemu-devel] [PATCH][RFC] Fix gcc 4.x warnings (-Wpointer-sign)

[Blue Swirl]

[-- Attachment #1: Type: text/plain, Size: 1145 bytes --]

On 9/13/08, Blue Swirl <blauwirbel@gmail.com> wrote:
> On 9/12/08, M. Warner Losh <imp@bsdimp.com> wrote:
>  > In message: <f43fc5580809121256i1d15c407vbebbfc8cd081cb43@mail.gmail.com>
>  >             "Blue Swirl" <blauwirbel@gmail.com> writes:
>  >  : GCC 4 enables -Wpointer-sign if -Wall is enabled. This trivial patch
>  >
>  > : suppresses the warnings. Comments? Objections?
>  >
>  >
>  > Don't you mean 'fix' the warnings, since this patch looks like it
>  >  tries to cast things to be correct...
>
>
> Well, some parts of the patch actually fix the types so that they are
>  more consistent, the casts just silence the warnings. A real fix would
>  be to change the types but then it would be incorrect to change the
>  signed types just to match the unsigned loadvm/savevm API.
>
>  Now that I think of it, the API could be extended with signed
>  versions. That would be best in terms of type safety. Maybe I'll try
>  that approach instead.

The first patch adds signed versions. This cleans up some parts
because casts are no longer needed.

The second patch fixes other warnings by making the types more
consistent, like before.

[-- Attachment #2: signed_savevm_api.diff --]
[-- Type: plain/text, Size: 18109 bytes --]

[-- Attachment #3: fix_signed_warnings.diff --]
[-- Type: plain/text, Size: 21174 bytes --]