On Mon, 12 Jan 2004, Post, Mark K wrote: > Sam, > > A question and a comment... > > Where did you get your kernel source from? I downloaded 2.4.24 from > kernel.org, and I don't see drivers/s390/char/hwc_rw.c in it at all.
I got it this morning (EST5EDT) from ftp://ftp.kernel.org/pub/linux/kernel/v2.4/linux-2.4.24.bz2 gpg --verify linux-2.4.24.tar.bz2.sign zips/linux-2.4.24.tar.bz2 gpg: Signature made Mon Jan 5 08:58:48 2004 EST using DSA key ID 517D0F0E gpg: Good signature from "Linux Kernel Archives Verification Key <[EMAIL PROTECTED]>" tar jtvf zips/linux-2.4.24.tar.bz2 | grep s390/char drwxr-xr-x marcelo/marcelo 0 2003-08-25 07:44:42 linux-2.4.24/drivers/s390/char/ -rw-r--r-- marcelo/marcelo 517 2003-08-25 07:44:42 linux-2.4.24/drivers/s390/char/ctrlchar.h -rw-r--r-- marcelo/marcelo 2021 2003-08-25 07:44:42 linux-2.4.24/drivers/s390/char/ctrlchar.c ... -rw-r--r-- marcelo/marcelo 49359 2002-11-28 18:53:14 linux-2.4.24/drivers/s390/char/hwc_rw.c -rw-r--r-- marcelo/marcelo 4502 2002-11-28 18:53:14 linux-2.4.24/drivers/s390/char/hwc.h ... > > The comment is that the problem with drivers/s390/char/ctrlchar.c has been > reported before, and a patch issued for the 2.4.21 kernel > (linux-2.4.21-s390-06-june2003.tar.gz). IBM has been having problems > getting their updates accepted into the official source tree, though, so it > didn't surprise me that the problem was still in 2.4.24. OK. Maybe cc marcelo? > > It made me think, though, whether the "experimental" patches IBM has for > 2.4.23 would apply cleanly to 2.4.24. So, I tried it, and the patches went > on without a peep. Cool, I thought. But, when I tried to compile that, the > ctrlchar.c problem was still there. Now, this _is_ a problem, because those > patches are supposed to be equivalent to the patches for the 2.4.21 kernel. > As I mentioned, the fix for ctrlchar.c was in the -06 patches for 2.4.21, > and that came out on October 31, 2003. So, it looks as though that fix got > missed somehow when the 2.4.23 patches were created. This raises the > question as to whether anything else might be missing as well, but doesn't > cause compilation errors. Maybe the config used didn't test that path. > I've copied the Boeblingen Lab email address on this note, so perhaps the > experimental patches can be updated to include this. > > Additionally, if you turn on the crypto support, you'll find a header file > completely missing: include/asm-s390/kmap_types.h > I compared the various architecture versions of the file, and most of them > seemed to be the same, with only a few comments being different. Copying > the i386 let things compile. Whether that _works_ or not, who knows? > I'm not running with crypto enabled so did not see this. > There was also a fix against include/asm-s390/ptrace.h that apparently never > made it into the tree: > --- linux-2.4.19/include/asm-s390/ptrace.h.bak Mon Mar 17 19:59:32 2003 > +++ linux-2.4.19/include/asm-s390/ptrace.h Mon Mar 17 20:00:05 2003 <snip/> > > I remember Gerhard Tonn saying that there still wasn't a satisfactory > official fix for this, though. Perhaps there still is not. That is likely worth further investigation. Thanks! > > Mark Post -- Sam Chessman chessman (a) tux.org First do what's necessary, then what's possible, finally the impossible.
