>> On Wed, Dec 3, 2008 at 19:42, Vorobiev Dmitri >> <dmitri.vorobiev@xxxxxxxxx> >> wrote: >>>> On Wed, 2008-12-03 at 18:52 +0100, Kay Sievers wrote: >>>>> On Wed, Dec 3, 2008 at 18:08, James Bottomley >>>>> <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> wrote: >>>>> > On Wed, 2008-12-03 at 18:24 +0200, Vorobiev Dmitri wrote: >>>>> >> > This patch fixes the following compilation warning: >>>>> >> > >>>>> >> > CC [M] drivers/scsi/sgiwd93.o >>>>> >> > drivers/scsi/sgiwd93.c:314: warning: initialization from >>>>> incompatible >>>>> >> > pointer type >>>>> >> >>>>> >> Any news about this one? I think this patch should go via >>>>> linux-scsi, >>>>> >> unless you would be insisting on pushing it via linux-mips, in >>>>> which >>>>> case >>>>> >> I'll politely bug Ralf about it. :) >>>>> > >>>>> > Looks OK for the local change. >>>>> > >>>>> > Globally, having driver->remove and platform_driver->remove return >>>>> int >>>>> > instead of void looks wrong. Particularly when the only use cases >>>>> are >>>>> > in drivers/base/ and they all ignore the return code. >>>>> > >>>>> > Greg and Kay ... shouldn't we simply redefine the return values for >>>>> the >>>>> > remove methods in these structures to return void (and thus match >>>>> the >>>>> > use case)? >>>>> >>>>> Aren't there many many drivers across the tree, using the "int >>>>> remove" >>>>> version? >>>> >>>> Yes ... since it's a function prototype. >>>> >>>> However, if drivers/base simply discards the return, it's a trap we >>>> shouldn't be setting. >>> >>> Hmmm, it does look like the return value is discarded, please see >>> drivers/base/dd.c::__device_release_driver() for details. >>> >>> Does this not deserve a good cleanup? >> >> Sure, it might be. If you want to patch hundreds of files, send >> patches to maintainers, patch drivers you can not even compile, we >> could do that. >> >> We are already in the middle of a ~400 files "struct device" bus_id >> conversion, and only very few maintainers respond to these patches. We >> also never got any reply to the SCSI bus_id patch we sent weeks ago. >> :) >> >> Even when it's "a good cleanup", with maintainers not responding, and >> supporting it, it's a real pain to change things like this. But, if >> you want to go ahead and do that, let us know. > > Well, I don't really want to look like a coward, but I guess this a good > project for Kernel Janitors, and I'm Cc:ing their mailing list now. ...and meanwhile, could you James please apply the original patch to silence the compiler warning until the void (*remove)() is introduced? Thanks, Dmitri
