Summary: | win4lin-sources 2.6.6 does not compile modules | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Karl Huysmans <karl.huysmans> |
Component: | [OLD] Core system | Assignee: | x86-kernel (DEPRECATED) <x86-kernel> |
Status: | RESOLVED FIXED | ||
Severity: | major | CC: | aoyu93, bugreports, phoenixreads, rodrigo |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 58048 | ||
Attachments: |
win4lin-sources-2.6.6-r2.ebuild
win4lin-sources-2.6.6-r2.ebuild (eclass version) win4lin-sources-2.6.6-r2.ebuild (eclass version) win4lin-sources-2.6.7-r1.ebuild win4lin-sources-2.6.7-r3 |
Description
Karl Huysmans
2004-06-12 11:01:07 UTC
Happens here too. I solved this by copying ~/linux/mki-adapter26-old/arch/i386/mki-adapter26 and its contents to ~/linux/arch/i386. Then built the kernel and all went well. Did this on 3 different machines, all ~x86 machines. Created attachment 33604 [details]
win4lin-sources-2.6.6-r2.ebuild
old style ebuild .. crufty etc. .. works but may have quirks
Created attachment 34067 [details]
win4lin-sources-2.6.6-r2.ebuild (eclass version)
The "problem" is the mki-adapter26_1_3_5.patch is not being applied properly.
This undoubtly has to do with the new eclass.
I made a simple hack which applies the patch correctly - the old way.
Yes, I know the patch is being applied twice, but the eclass is doing it
incorrectly.
Side note: eclasses are "kewl", imagine them as default "modules" with
inheritance
Created attachment 34068 [details]
win4lin-sources-2.6.6-r2.ebuild (eclass version)
The orginal win4lin-sources-2.6.6-r1.ebuild does not use the $MKI_VERSION
variable in the SRC_URI variable - fixed.
I've just bumped win4lin-sources to 2.6.7 and added a new mki-adapter patch, so I'm closing this bug. Please reopen is the problem still occurs. Created attachment 34705 [details, diff]
win4lin-sources-2.6.7-r1.ebuild
2.6.7-r1 still has the same problem. I modified the patch so that
kernel-2.eclass can just handle it properly. The patch could instead be removed
from UNIPATCH_LIST and applied from the ebuild as in 2.6.6-r2, but I think it
should be one or the other, as the 2.6.6-r2 ebuild seems to apply it twice.
Oops, that was an accident. That's not a patch, of course. Sorry. Created attachment 35356 [details]
win4lin-sources-2.6.7-r3
this form of the win4lin-sources-2.6.7-r3 ebuild works .. the other "sed" style
does not(commented out in this ebuild) ..
of course it double patches but the mki-adapter26-old sub-directory at the top
level is ignored when building the kernel ..
Just to leave a note: I have just emerged the 2.6.7-win4lin-r2 sources and it still doesn't create the mki-adapter module by itself. I aplied the mki-adapter patch manually and everything got built correctly. I believe this bug should be reopened. Or should I create a new one? to all concern: patiently wait... John Mylchreest said he was going to modify the eclass to allow explict patch levels - still waiting :) It will solve this problem period. {patch_filename}:{pathlevel} All that will have to be changed is ${DISTDIR}/mki-adapter26_${MKI_VERSION}.patch to ${DISTDIR}/mki-adapter26_${MKI_VERSION}.patch:1 Sidenote: Doesn't hurt to talk to the developer(s) instead of making numerious workaround, no? Also please note: 1) the problem has not been acknowledge (see #6), it was not the patch itself but HOW the patch is being applied. This is due to using the new kernel-2.eclass. 2) nor has any solution been accepted (see win4lin-sources-2.6.7-r2, no change, not even a hack.) #10 points out the problem still exists. Any listening? Use the hack until John Mylchreest implements the changes and everyone will be happy. Commenting on #11 prooposed hack: It didn't work for me. I had to make that change in /usr/portage/sys-kernel/win4lin-sources/win4lin-sources-2.6.7-r2.ebuild file, right? Well, I'm still apllying mki-adapter patch manually... I am glad to hear that a proper fix is underway. I would like to ask what the problem you had with my sed hack was, though. Are you sure that you modified the UNIPATCH_LIST as well as adding a new src_unpack? I changed ${DISTDIR}/mki-adapter26_${MKI_VERSION}.patch to ${WORKDIR}/mki-adapter26_${MKI_VERSION}.patch. I ask because I just tried it myself with 2.6.7-r3, and it built the module as it should. I am curious about what I did differently. No, I didn't. I got the present 2.6.7-r2 ebuild and just made the ":1" change you mentioned. I think I got it now: I have to get the ebuild from #9 and make the modification you proposed in #11, is that right? I won't be able to try it now as I'm already late for airport but please leave a note to let me know exactly what should I try and I will do it tomorow. To answer #14. :1 is the future fix, which I am patiently waiting for win4lin-sources-2.6.7-r3 is most current ebuild USING the hack and should work without modification. The commented 'src_unpack' section is the sed hack. The 'src_unpack' section at the bottom is the "old-way" hack. Any questions. :) I'm sorry, that was not at all clear of me. My previous question was in reply to comment #9. win4lin-sources-2.6.7-r3.ebuild isn't working either, as far as I can tell. It's choking on the lack of "win4lin-sources.CAN-2004-0497.patch". After I removed it from UNIPATCH_LIST, the ebuild choke on the lack of "win4lin-sources-2.6.IPTables-RDoS.patch". I believe "win4lin-sources-2.6.ProcPerms.patch" would also be a problem as none of these three files are listed in SRC_URI. Or am I missing something? There are two source of files, one which downloads files to /distfile directory and other is in /files directory with the ebuild - this is where the patches are hiding. So make a link from (will save you future trouble) /usr/portage/sys-kernel/win4lin-sources/files to {location of local portage tree}/sys-kernel/win4lin-sources/files The variable ${FILESDIR} is relative to the ebuild file. Extra info on win4lin-sources-2.6.7-r3.ebuild. I was making a mess with win4lin-sources-2.6.7-r3. I thought it wasn't yet available on regular portage so I downloaded it from this page. This is the reason for the "files not available" problems I reported on #17. Having clarified this point let me report my experiences with the real win4lin-sources-2.6.7-r3.ebuild: the mki-adapter module is still not created. I still have to manually (re)aply the mki-adapter patch. As far as I can tell, win4lin-sources-2.6.7-r3.ebuild is no-go. Enough already... http://bugs.gentoo.org/show_bug.cgi?id=58008 1) commented out the patch in UNIPATCH_LIST so it is not applied twiced 2) added a line so that if placed in the local portage tree it will link to the patch files - no more missing files 3) make a new report so all can see. Would you believe other people are having problems with win4lin not working? I have just tested win4lin-sources-2.6.7-r4.ebuild from http://bugs.gentoo.org/show_bug.cgi?id=58008. It works!! I mean, mki-adapter module gets built. Despite being refered to as a hack I think win4lin-sources-2.6.7-r4.ebuild should be published as the present stable win4lin-sources ebuild doesn't work with win4lin (no mki_adapter module). And them this bug could be considered RESOLVED/FIXED. Ops, it's already RESOLVED/FIXED ;) |