On Mon, 2009-05-18 at 17:03 +0100, Alan Jenkins wrote: > On 5/18/09, Andreas Robinson <andr345@xxxxxxxxx> wrote: > > So, when asked to load a file that is not an elf-file, what should modprobe > > do? > > > > 1. fail silently if (error == warn && !verbose) or > > 2. always fail noisily, same as if e.g the file didn't exist. > > > > Up to this point, case 1 is true. > > After applying the patch, case 2 would be true, if the hack didn't revert to > > the old behvior. > > > > I'm not sure what the right thing to do is here. > > For what it's worth, 2) sounds fine to me. I concur. There is never a situation today in which that would be an issue, and short of a fairly substantial module format change, it's never going to be an issue either. > I guess it also means corrupted module files won't show up in the > kernel log. (Which was theoretically convenient for early boot or > headless systems). But I don't think it's worth worrying about. Nah. I mean, it would never be good to rely on such kernel logs to catch every possibility anyway. > Apart from corrupted files, it sounds like this would also apply when > trying to load a compressed module without enabling zlib? I think > it's right to be noisy in these cases, and it will make more sense if > the error comes from userspace instead of from the kernel. Indeed. And that's fine again. Appreciate very much the patch series, especially patch 4. I had to wait on testing on PPC64 until this morning (have a replacement network switch now), which is why you didn't see the release yet, however I'm about to push an updated tree so that we don't have to wait for that. I am not willing to push a release with the ELF changes without checking LE/BE on 32/64-bit targets properly this morning. Somewhat related, I had someone moan about the binary format change (not being able to build with newer version and use with old) affecting cross-compiling on ARM (again) over the weekend - another reason I really don't just want to test regular x86 systems. Fortunately I now have dedicated test hardware at home attached to new IP power switches so I can shortly automate run_tests. Jon. -- To unsubscribe from this list: send the line "unsubscribe linux-modules" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
