Re: question on common error-handling idiom

[Jesper Juhl <juhl-lkml@dif.dk>]

On Tue, 2 Nov 2004, Jan Engelhardt wrote:

> >There are some places that do
> >
> >err = -SOMEERROR;
> >if (some_error)
> >	goto out;
> >if (some_other_error)
> >	goto out;
> >if (another_error)
> >	goto out;
> >
> >Let's see what other people think :)
> 
> err = -ESOME;
> if(some_error || some_other_error || another_error) {
> 	goto out;
> }
> 
> Best.
> 
Agreed, but that would potentially make something like the following (from 
fs/binfmt_elf.c::load_elf_binary) quite unreadable :

if (loc->elf_ex.e_type != ET_EXEC && loc->elf_ex.e_type != ET_DYN)
	goto out;
if (!elf_check_arch(&loc->elf_ex))
	goto out;
if (!bprm->file->f_op||!bprm->file->f_op->mmap)
	goto out;


But that's not even the most interresting case, the interresting one is 
the one that does

err = -ERR;
if (foo)
	goto out;

err = -ERROR;
if (bar)
	goto out;

where there's the potential to save an instruction by moving the err= into 
the error path.
But as Oliver Neukum pointed out gcc may not be smart enough for that to 
actually generate the best code.

Has anyone taken a look at what recent gcc's actually do with different 
variations of the constructs mentioned in this thread?


--
Jesper Juhl