warning from gcc version 4.6.0 20100416
in [Kernel]
|
Prev: [PATCH] drivers/usb/gadget/langwell_udc.c: printk needs a (unsigned long long) cast for a dma_t
Next: linux-next: Tree for June 11
From: Andi Kleen on 14 Jun 2010 04:00 On Sun, Jun 13, 2010 at 09:17:41PM -0400, Bill Davidsen wrote: > Justin P. Mattock wrote: >> o.k. andi, >> >> here is the rest of the warnings that >> I see when compiling the kernel >> >> I can try and create some patches for >> this(hopefully!!) >> > There is no great solution to this, in a fair number of cases the fix would > slow the code or make it harder to read, so some of these probably don't Sorry that's wrong: the optimizer will generate the same code anyways as if the unused variable was not there because it eliminates unused variables. So fixing this cannot make code slower. I also don't see how unused variables make the code easier to read. The only difficult case sometimes is with #ifdef code, that has to be handled case by case. One elegant solution is to replace the ifdef code with an inline. > want a fix. Of course some clearly are errors, so you are doing something All warnings should be fixed, I only left those in that are real code bugs if I couldn't fix the code. Kernel builds are expected to be relatively warning free so that you can easily spot new warnings. But eventually someone who knows the code better has to fix that bug. --Andi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Bill Davidsen on 14 Jun 2010 11:00
Andi Kleen wrote: > On Sun, Jun 13, 2010 at 09:17:41PM -0400, Bill Davidsen wrote: > >> Justin P. Mattock wrote: >> >>> o.k. andi, >>> >>> here is the rest of the warnings that >>> I see when compiling the kernel >>> >>> I can try and create some patches for >>> this(hopefully!!) >>> >>> >> There is no great solution to this, in a fair number of cases the fix would >> slow the code or make it harder to read, so some of these probably don't >> > > Sorry that's wrong: the optimizer will generate the same > code anyways as if the unused variable was not there > because it eliminates unused variables. > > If one puts an 'if' around the variable set (I've seen it) the test may well take longer than the assign, assuming that there's a case where the assign is done. > So fixing this cannot make code slower. > > I also don't see how unused variables make the code easier > to read. > > No, my point was that putting a bunch of ifdef statements in to avoid the warning will make the code harder to read. > The only difficult case sometimes is with #ifdef code, > that has to be handled case by case. One elegant solution > is to replace the ifdef code with an inline. > > >> want a fix. Of course some clearly are errors, so you are doing something >> > > All warnings should be fixed, I only left those in that > are real code bugs if I couldn't fix the code. > > Kernel builds are expected to be relatively warning free > so that you can easily spot new warnings. > > But eventually someone who knows the code better has to > fix that bug. > > --Andi > > -- Bill Davidsen <davidsen(a)tmr.com> "We can't solve today's problems by using the same thinking we used in creating them." - Einstein -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |