Re: RFC: util-linux's FTBFS
On Wed, Jul 22, 2009 at 01:39:25AM +0200, Cyril Brulebois wrote:
> Heya,
>
> new util-linux is a prerequisite for the whole hal thing. I've just
> downloaded and built it from git.debian.org[1], and I encountered the
> following problem with ENOMEDIUM, that seems to be Linux-specific. I'm
> attaching a proposed patch. I'm quoting the whole context:
> | /* ask kernel developers why we need such ugly open() method... */
> | static int
> | open_device(const char *devname)
> | {
> | int retries = 0;
> |
> | do {
> | int fd = open(devname, O_RDONLY);
> | if (fd >= 0)
> | return fd;
> | #ifdef ENOMEDIUM
> | if (errno != ENOMEDIUM)
> | break;
> | #endif
> | if (retries >= CRDOM_NOMEDIUM_RETRIES)
> | break;
> | ++retries;
> | sleep(3);
> | } while(1);
> |
> | return -1;
> | }
>
> Do you people think of a better patch? *ding* Reading it again,
> I guess it'd be better to just add “#else break” in there, no?
I have done a test. The open of a CD-ROM device without medium loaded on
GNU/kFreeBSD is successful. Then reading from it returns -EINVAL.
So it looks like a #else break is the right solution.
> 1. git://git.debian.org/users/lamont/util-linux.git [master]
> ./autogen.sh && debuild -b
> It is supposed to reach experimental first.
>
>
> After the build, the symbols file shows missing uuid_{,un}pack:
> | - uuid_pack@UUID_1.0 2.16-1
> | +#MISSING: 2.16-2# uuid_pack@UUID_1.0 2.16-1
> | - uuid_unpack@UUID_1.0 2.16-1
> | +#MISSING: 2.16-2# uuid_unpack@UUID_1.0 2.16-1
>
> I'll start and figure out where that comes from later tonight.
>
That's a packaging bug as the problem is the same on GNU/Linux.
Cheers,
Aurelien
--
Aurelien Jarno GPG: 1024D/F1BCDB73
aurelien@aurel32.net http://www.aurel32.net
Reply to: