2.19 -enable-targets=all one error in m68k-dis.c due to longjmp, easily fixed with volatile

[Nick Clifton]

Hi Alan, Hi Jay,

> In this particular case I'm quite sure the warning is
> false, so it isn't a good idea to slow down this function by making
> "info" volatile.

Agreed.

> Perhaps we should turn off -Werror in the opcodes/
> directory instead as there isn't a way to turn off just the longjmp
> without turning off a bunch of other errors.

I disagree.  I think that we should keep -Werror as it does help to find 
real bugs.

The real problem it seems to me is that m68k-dis.c is using setjmp / 
longjmp when there really is no need for it.  It was just to save the 
author from writing a whole lot of error value handling code.  So I have 
gone ahead and written that code.

Jay - would you care to try out the attached patch and let me know if it 
works for you ?  Assuming that it does I will check the patch into the 
branch and the mainline.

Cheers
   Nick



-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: m68k-dis.c.patch
URL: <https://sourceware.org/pipermail/binutils/attachments/20081124/3cb07922/attachment.ksh>