Re: [Xen-devel] [PATCH for-4.8] ipxe: update to newer commit [and 1 more messages]

[Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>]

On 10/12/2016 07:00 AM, Ian Jackson wrote:
> Ian Jackson writes ("Re: [Xen-devel] [PATCH for-4.8] ipxe: update to newer 
> commit"):
>> That was eventually done.  I'm not sure exactly when the change was
>> made.  Does gcc -Wno-foo work properly on all the gcc's we care about ?
> Jan Beulich writes ("Re: [Xen-devel] [PATCH for-4.8] ipxe: update to newer 
> commit"):
>> Just fyi I have run into an issue with -Wno-override-init use in Linux
>> 4.8 on gcc 4.1.x, so what you say doesn't appear to hold for all gcc
>> versions we permit to be used.
> Well, that answers my question above.
>
> I think the right approach is to:
>
>  * Test -Wno-this-is-not-a-warning-option.  If gcc accepts it,
>    add -Wno-something to disable the nonnull check·

Back compatibility is in fact not a problem. These options would only be
passed on when gcc6+ is used

>
>  * Review the misleading indentations and if there are only a few, fix
>    them in a downstream patch.  Or, if there are many, decide to
>    tolerate them.

There are more warnings than just indentation and nonnull checks:
    -Wno-nonnull-compare
    -Wno-unused-const-variable
    -Wno-misleading-indentation
    -Wno-shift-negative-value
    -Wno-array-bounds

(The last two flagged actual bugs that have been fixed upstream).

Some of the warnings can be addressed by backporting upstream patches
but there are a few for which backporting will involve much more code
movement than fixing the code ourselves.

-boris


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel