Old posts from the morph-sources thread

[yardbird]

morph-sources includes as, which is similer to ac, but doesn't have the drivers your looking for

I wonder why is it this way... I've been struggling with morph/ac/as to isolate it82xx from ac and I concluded it would be better to include to whole -ac patch because of the many nice fixes/improvements it contains that do not appear on -as, really.
Besides, I'm very much confident on the work of Alan so I trust the -ac patchset will not yield dangerous unstabilities.

So... is it possible/reasonable to morph-sources include ac instead of as? Or both?

What do you people think?

Some of previous morph-sources were based on -ac indeed. I chose to switch to -as because -ac didn't have any feature urgently demanded by the community, and -as seemed a safer choice since it is bugfix-only. However I can switch back to -ac if there is enough interest in it, I trust Alan Cox too

Let me know what you folks think.

Oh, BTW, an unstable morph based on 2.6.11-rc4 will be out tomorrow. I managed to mantain all the features of morph-sources 2.6.10, and I'm getting ready to switch to 2.6.11. But the 2.6.10 tree will live on and will be acitvely worked on at least until 2.6.11 final is out. Cheers!

EDIT: typos.

[rel]

Hi, wonder if it is possible to reverse the patches I don't use.
I always tend to think that added code touches and influences other code even when not using
the added code, like reiser4. But this even when not compiled in impacts the behaviour of the kernel.

So I won't use reiser4, xbox, bluetooth. I know it is possible to reverse, but this has to be in order? Doesn't it?
I'am bad at this and never working out, getting hunks failed etc. :/
It would be somthing great to specify te patches at ebuild level, no one big monolithic patch. But loose patches
and apply these at useflag level? Just what the user needs.

morphastic

[yardbird]

Hi, wonder if it is possible to reverse the patches I don't use.
I always tend to think that added code touches and influences other code even when not using
the added code, like reiser4. But this even when not compiled in impacts the behaviour of the kernel.

So I won't use reiser4, xbox, bluetooth. I know it is possible to reverse, but this has to be in order? Doesn't it?
I'am bad at this and never working out, getting hunks failed etc. :/
It would be somthing great to specify te patches at ebuild level, no one big monolithic patch. But loose patches
and apply these at useflag level? Just what the user needs.

morphastic

Well, for some patches this is doable. For example the xbox and bluetooth patches you mentioned are not in the "core" morph-sources, rather they are specified by USE flags ("xbox" and "updated-bluetooth"). Unfortunately some patches interact with each other and it becomes difficult to apply them separately. For example the reiser4 patch changes some files upon which other patches depend upon, and these patches has to be adapted to cope with this. But if reiser4 is not applied the other patches get some rejects. In some cases this problem is solvable (staircase and cfq-ts, for example), in others it is not.

Anyway, it could be a good idea to pull out the patches that does not interact with other patches and provide them as USE flags. I'll think about it.

[pilx]

Hey, I'm seeing sistematic inotify hangups that brings down the whole system consistently
Don't know where to post the bug so here is a representative piece of log...

Code: Select all

Feb 15 02:53:02 tulkas Unable to handle kernel NULL pointer dereference at virtual address 00000008
Feb 15 02:53:02 tulkas printing eip:
Feb 15 02:53:02 tulkas c02e5e24
Feb 15 02:53:02 tulkas *pde = 00000000
Feb 15 02:53:02 tulkas Oops: 0000 [#1]
Feb 15 02:53:02 tulkas PREEMPT
Feb 15 02:53:02 tulkas Modules linked in: nls_iso8859_1 isofs zlib_inflate snd_pcm_oss snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device parport_pc parport 8250_pnp 8250 serial_core floppy nvidia ohci1394 ieee1394 tuner saa7134 video_buf v4l2_common v4l1_compat ir_common videodev usblp snd_intel8x0 snd_ac97_codec snd_pcm snd_timer snd_page_alloc ehci_hcd ohci_hcd i2c_nforce2 i2c_core usbcore snd_mixer_oss snd soundcore
Feb 15 02:53:02 tulkas CPU:    0
Feb 15 02:53:02 tulkas EIP:    0f20:[<c02e5e24>]    Tainted: P      VLI
Feb 15 02:53:02 tulkas EFLAGS: 00210287   (2.6.10-nitro4)
Feb 15 02:53:02 tulkas EIP is at inotify_dev_queue_event+0x74/0x170
Feb 15 02:53:02 tulkas eax: 00000000   ebx: d5094460   ecx: 00000100   edx: c19328c8
Feb 15 02:53:02 tulkas esi: 00000100   edi: cf944b58   ebp: 00000000   esp: c3fa1ef0
Feb 15 02:53:02 tulkas ds: 0f3b   es: 0f3b   ss: 0f28
Feb 15 02:53:02 tulkas Process emerge (pid: 16588, threadinfo=c3fa0000 task=c5e9c540)
Feb 15 02:53:02 tulkas Stack: 00000f82 c5e9c698 c3fa0000 ffffffff cf944b58 00000000 c19328c8 c3fa0000
Feb 15 02:53:02 tulkas c19328c8 cf944b58 00000000 c02e6546 d5094460 c19328c8 00000100 00000000
Feb 15 02:53:02 tulkas cf944b58 00000000 cf944af4 da073458 c3fa1f70 c016a838 da073458 00000100
Feb 15 02:53:02 tulkas Call Trace:
Feb 15 02:53:02 tulkas [<c02e6546>] inotify_inode_queue_event+0x56/0x90
Feb 15 02:53:02 tulkas [<c016a838>] vfs_rmdir+0x138/0x220
Feb 15 02:53:02 tulkas [<c016a9b8>] sys_rmdir+0x98/0x100
Feb 15 02:53:02 tulkas [<c01599f9>] filp_close+0x59/0x90
Feb 15 02:53:02 tulkas [<c01031f3>] sysenter_past_esp+0x58/0x79
Feb 15 02:53:02 tulkas Code: 0f 87 a8 00 00 00 0f 84 b6 00 00 00 81 f9 00 20 00 00 74 33 81 f9 00 80 00 00 74 2b 8b 54 24 18 89 ce 8b 42 08 8b 80 0c 01 00 00 <23> 70 08 85 f6 747c 81 f9 00 80 00 00 74 0d 89 c8 23 42 04 85
Feb 15 02:53:02 tulkas <6>note: emerge[16588] exited with preempt_count 1
Feb 15 02:53:02 tulkas scheduling while atomic: emerge/0x10000001/16588
Feb 15 02:53:02 tulkas [<c040da11>] schedule+0x491/0x4a0
Feb 15 02:53:02 tulkas [<c01499a0>] unmap_page_range+0x70/0x90
Feb 15 02:53:02 tulkas [<c0118b1c>] cond_resched_lock+0x2c/0x50
Feb 15 02:53:02 tulkas [<c0149b1a>] unmap_vmas+0x15a/0x1d0
Feb 15 02:53:02 tulkas [<c014e3a3>] exit_mmap+0x83/0x160
Feb 15 02:53:02 tulkas [<c01193af>] mmput+0x3f/0x100
Feb 15 02:53:02 tulkas [<c011dba8>] do_exit+0xd8/0x4d0
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c010449b>] die+0x18b/0x190
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c011ba87>] printk+0x17/0x20
Feb 15 02:53:02 tulkas [<c0115c6d>] do_page_fault+0x2bd/0x5b8
Feb 15 02:53:02 tulkas [<c0117597>] slice+0x27/0x40
Feb 15 02:53:02 tulkas [<c01175f4>] effective_prio+0x44/0xe0
Feb 15 02:53:02 tulkas [<c0117816>] activate_task+0x56/0x80
Feb 15 02:53:02 tulkas [<c011749b>] __activate_task+0x1b/0x30
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c0103cd3>] error_code+0x2b/0x30
Feb 15 02:53:02 tulkas [<c02e5e24>] inotify_dev_queue_event+0x74/0x170
Feb 15 02:53:02 tulkas [<c02e6546>] inotify_inode_queue_event+0x56/0x90
Feb 15 02:53:02 tulkas [<c016a838>] vfs_rmdir+0x138/0x220
Feb 15 02:53:02 tulkas [<c016a9b8>] sys_rmdir+0x98/0x100
Feb 15 02:53:02 tulkas [<c01599f9>] filp_close+0x59/0x90
Feb 15 02:53:02 tulkas [<c01031f3>] sysenter_past_esp+0x58/0x79
Feb 15 02:53:02 tulkas scheduling while atomic: emerge/0x10000001/16588
Feb 15 02:53:02 tulkas [<c040da11>] schedule+0x491/0x4a0
Feb 15 02:53:02 tulkas [<c01499a0>] unmap_page_range+0x70/0x90
Feb 15 02:53:02 tulkas [<c0118b1c>] cond_resched_lock+0x2c/0x50
Feb 15 02:53:02 tulkas [<c0149b1a>] unmap_vmas+0x15a/0x1d0
Feb 15 02:53:02 tulkas [<c014e3a3>] exit_mmap+0x83/0x160
Feb 15 02:53:02 tulkas [<c01193af>] mmput+0x3f/0x100
Feb 15 02:53:02 tulkas [<c011dba8>] do_exit+0xd8/0x4d0
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c010449b>] die+0x18b/0x190
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c011ba87>] printk+0x17/0x20
Feb 15 02:53:02 tulkas [<c0115c6d>] do_page_fault+0x2bd/0x5b8
Feb 15 02:53:02 tulkas [<c0117597>] slice+0x27/0x40
Feb 15 02:53:02 tulkas [<c01175f4>] effective_prio+0x44/0xe0
Feb 15 02:53:02 tulkas [<c0117816>] activate_task+0x56/0x80
Feb 15 02:53:02 tulkas [<c011749b>] __activate_task+0x1b/0x30
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c0103cd3>] error_code+0x2b/0x30
Feb 15 02:53:02 tulkas [<c02e5e24>] inotify_dev_queue_event+0x74/0x170
Feb 15 02:53:02 tulkas [<c02e6546>] inotify_inode_queue_event+0x56/0x90
Feb 15 02:53:02 tulkas [<c016a838>] vfs_rmdir+0x138/0x220
Feb 15 02:53:02 tulkas [<c016a9b8>] sys_rmdir+0x98/0x100
Feb 15 02:53:02 tulkas [<c01599f9>] filp_close+0x59/0x90
Feb 15 02:53:02 tulkas [<c01031f3>] sysenter_past_esp+0x58/0x79

This happens with morph-sources-2.6.10-r23 and with nitro-sources-2.6.10-r4 both with inotify on and gamin running.

[m0sia]

i upgraded from morph20 to morph23 with $make oldconfig. after rebooting i have error. something like:
"vfs: kernel panic
sda2 not found
use root= to set your root partition"

wtf?

[rel]

Hey, I'm seeing sistematic inotify hangups that brings down the whole system consistently
Don't know where to post the bug so here is a representative piece of log...

Code: Select all

Feb 15 02:53:02 tulkas Unable to handle kernel NULL pointer dereference at virtual address 00000008
>>
Feb 15 02:53:02 tulkas [<c0115c6d>] do_page_fault+0x2bd/0x5b8
Feb 15 02:53:02 tulkas [<c0117597>] slice+0x27/0x40
Feb 15 02:53:02 tulkas [<c01175f4>] effective_prio+0x44/0xe0
Feb 15 02:53:02 tulkas [<c0117816>] activate_task+0x56/0x80
Feb 15 02:53:02 tulkas [<c011749b>] __activate_task+0x1b/0x30
Feb 15 02:53:02 tulkas [<c01159b0>] do_page_fault+0x0/0x5b8
Feb 15 02:53:02 tulkas [<c0103cd3>] error_code+0x2b/0x30
Feb 15 02:53:02 tulkas [<c02e5e24>] inotify_dev_queue_event+0x74/0x170
Feb 15 02:53:02 tulkas [<c02e6546>] inotify_inode_queue_event+0x56/0x90
Feb 15 02:53:02 tulkas [<c016a838>] vfs_rmdir+0x138/0x220
Feb 15 02:53:02 tulkas [<c016a9b8>] sys_rmdir+0x98/0x100
Feb 15 02:53:02 tulkas [<c01599f9>] filp_close+0x59/0x90
Feb 15 02:53:02 tulkas [<c01031f3>] sysenter_past_esp+0x58/0x79

This happens with morph-sources-2.6.10-r23 and with nitro-sources-2.6.10-r4 both with inotify on and gamin running.

Had the same issue with a custom patched kernel, using the latest inotify .18 and gamin.
Now with morph22 these are over. BUT even if gamin is running, it seems not to do its job.
Meaning: in Nautilus I have to hit reload to see changes on the fs. What would be causing this?
Wonder if fam would have this issues with inotify.

rel

[rel]

Hi, wonder if it is possible to reverse the patches I don't use.
I always tend to think that added code touches and influences other code even when not using
the added code, like reiser4. But this even when not compiled in impacts the behaviour of the kernel.

So I won't use reiser4, xbox, bluetooth. I know it is possible to reverse, but this has to be in order? Doesn't it?
I'am bad at this and never working out, getting hunks failed etc. :/
It would be somthing great to specify te patches at ebuild level, no one big monolithic patch. But loose patches
and apply these at useflag level? Just what the user needs.

morphastic

Well, for some patches this is doable. For example the xbox and bluetooth patches you mentioned are not in the "core" morph-sources, rather they are specified by USE flags ("xbox" and "updated-bluetooth"). Unfortunately some patches interact with each other and it becomes difficult to apply them separately. For example the reiser4 patch changes some files upon which other patches depend upon, and these patches has to be adapted to cope with this. But if reiser4 is not applied the other patches get some rejects. In some cases this problem is solvable (staircase and cfq-ts, for example), in others it is not.

Anyway, it could be a good idea to pull out the patches that does not interact with other patches and provide them as USE flags. I'll think about it.

Yeah, It's indeed hard to modularize most of the patches. Maybe an idea to have multiple ebuilds (or maybe a bad idea) of the same version, where they
have a base. one with reiser4 is applied, on whitch useflagged patches can apply. And one without where other useflagged patches can be left out. Am I talking
bully here?

Hmm, still wonder if code of patches can influence the kernel in a bad sence even if those patches are not selected in .config.

anyhow, morp22 is running happily here.


rel.

[rel]

i upgraded from morph20 to morph23 with $make oldconfig. after rebooting i have error. something like:
"vfs: kernel panic
sda2 not found
use root= to set your root partition"

wtf?

Guess you made a typo setting up grub?


rel.

[pilx]

Had the same issue with a custom patched kernel, using the latest inotify .18 and gamin.
Now with morph22 these are over. BUT even if gamin is running, it seems not to do its job.
Meaning: in Nautilus I have to hit reload to see changes on the fs. What would be causing this?
Wonder if fam would have this issues with inotify.
rel

AFAIK fam has no knowledge of inotify, so it would still be using dnotify.
Gamin is working here, using inotify, until hangup like above ocurrs; "is working" means _KDE_ (konqueror & krusader) shows filesystem changes automatically.
I can reproduce the hangup opening a non empty folder with konqueror and then deleting that folder from the shell... then rm process hangs and gam_server starts eating up CPU and the system log start showing entries like above (continuosly), then if I pkill -9 gam_server the whole system stops responding.

[HardenCoonor]

Hi,

I also get a kernel crash with morph23, but these happen only when I power down or reboot, at the end, after "remounting filesytems readonly", so this is no problem for me. This is maybe because i always activate reiser4 with 4k-stacks, in the Kconfig.reiser4.

BTW. concerning the ebuild, isn't it better to make the mppc-patch an optional download? I always do that myself, but doing it everytime with every new morph is a bit annoying, for the same reason morph is currently my favourite patchset, because of the frequent and most of the time working releases.

It would look like this

Code: Select all

 mppc-mppe? ( ) 

in SRC URI (Sorry, i am on windows, so i don't have the ebuild open, since the new release is not out yet.)

Meanwhile, thanks, yardbird, for your continuing effort to enrich this world with another good kernel.

[yardbird]

Hi,

to those of you who are experiencing problems with inotify and gamin, have you updated to the latest gamin? inotify 0.18 & 0.19 change the internals of inotify and gamin has to be patched to cope with this. I recommend using the latest gamin (0.23, should be...) with this patch applied. Just start emerge gamin, hit ctrl+Z after source is unpacked, patch it (it is in /var/tmp/portage/gamin/work/...) and continue the installation with "fg".

HardenCoonor: I've added your suggestion to my TODO list

rel: I'll try to modularize morph-sources as much as possible. I want to keep things simple though, so I'll keep providing a core patch with addons.

[yardbird]

I've released a pre-release version based on 2.6.11-rc4. All the main features of morph-sources are there, head to the website for details.

I'm also working to merge -ac in place of -as in the next 2.6.10 release. There will also be a softwaresuspend update, I'll post as soon as it is ready.

Cheers!

[rel]

Hi,

to those of you who are experiencing problems with inotify and gamin, have you updated to the latest gamin? inotify 0.18 & 0.19 change the internals of inotify and gamin has to be patched to cope with this. I recommend using the latest gamin (0.23, should be...) with this patch applied. Just start emerge gamin, hit ctrl+Z after source is unpacked, patch it (it is in /var/tmp/portage/gamin/work/...) and continue the installation with "fg".

HardenCoonor: I've added your suggestion to my TODO list

rel: I'll try to modularize morph-sources as much as possible. I want to keep things simple though, so I'll keep providing a core patch with addons.

And I have read about it on a blog issueing the inotfiy .18 radical change that breaks things (beagle). Didn't realize this also applied to gamin.
Installed gamin 0.23 and its working again.

Thanks for considering the modularization of your patchset. It would be really nice to put a bit of "freedom of choice" in it

rel.

[HardenCoonor]

@ yardbird: Thanks, but there is no need to give it a high priority, a nice +10 suffices, so you can release good kernels and still have a real life (is there any outside linux ?).
New versions might help with solving the issue with morph-sources and the kernel panic i always get when powering down or restarting my system. It seems to happen right after umounting, before remounting readonly, but there seem to be no data losses. I use a couple of reiser4 (/, /tmp, /home) and 2 ext3 partitions.

[_sys/sid]

Where is the ebuild for morph 2.6.11_rc4 ?

[yoyo]

Where is the ebuild for morph 2.6.11_rc4 ?

On morph website (http://morph-sources.sourceforge.net/) you can read :

There's no ebuild available, and there won't be until 2.6.11 goes stable. So if you want to try this out, grab the standalone patch from the download section.

You have to apply the patch yourself on the latest development-sources (2.6.11-rc4).
It works great for me :

Code: Select all

% uname -r
2.6.11-rc4-morph0

[pilx]

I also get a kernel crash with morph23, but these happen only when I power down or reboot, at the end, after "remounting filesytems readonly", so this is no problem for me. This is maybe because i always activate reiser4 with 4k-stacks, in the Kconfig.reiser4.

I'm seeing this too... and not using 4k stacks.

I'm also working to merge -ac in place of -as in the next 2.6.10 release. There will also be a softwaresuspend update, I'll post as soon as it is ready.

Hey, Great news!

I'll try the inotify/gamin updates patches when I get back to home.

[yardbird]

2.6.10-morph24 is out. Many updates, mainly the latest softwaresuspend patch and the -ac12 patch as requested. Also there are some more fixes for reiser4 on 64bit archs. More details on the website.

-------

HardenCoonor and Pilx: would you please try this out to see if the shutdown problem is gone? I experienced it once yesterday on my home server (even if on the laptop it never appeared). Can you reproduce it always?

Bye

[JBerro]

Hi yarbird, hi the morph-sources community,

Unlike the 22 and 23 The morph-sources24 no longer reboot when booting on AMD64 platform.
There is a trivial compilaton error as the file <asm/kbd.h> included in include /linux/kbh.h doen't exist on the AMD64 platform.
Just commenting this inlude solved the issue, not a big deal.
So i am quiet happy as i can now use the lastest morph-sources.
Using USE to define some optional patches is really a good idea.
I also support the focus on the stability.
I will now focus on trying reiser4.
This time i expect not to mess the entire system.

[yardbird]

Hi yarbird, hi the morph-sources community,

Unlike the 22 and 23 The morph-sources24 no longer reboot when booting on AMD64 platform.
There is a trivial compilaton error as the file <asm/kbd.h> included in include /linux/kbh.h doen't exist on the AMD64 platform. Just commenting this inlude solved the issue, not a big deal.

Thanks for the report, I'll add this fix in the next release. Glad to hear your problems got solved

So i am quiet happy as i can now use the lastest morph-sources.
Using USE to define some optional patches is really a good idea.
I also support the focus on the stability.
I will now focus on trying reiser4.
This time i expect not to mess the entire system.

Nice! Just be careful and do backups, reiser4 is not completely stable on 64 bit archs yet. Also you may need a very new version of reiser4tools, which should be out in a day or two (at least a release candidate should be out...). Alternatively you can patch version 1.0.3 with the patches you can find here.

Have a nice day!

[JBerro]

Hi yardbird,

I didn't lost any data as my home partition was still in reiserfs and additionaly i have daily backup of them (36 GB of picture...). I am aventurous but not crasy.
For now i will just move /tmp to reiser4, so if it breaks no issue.
Just one question for my understanding: as far as i know this is no way to chose the kernel stack size in the kernel config (at least on the AMD64 platform).
The parameter is still there but not set and not available in xconfig and not set, so my assumption was a 8k stack size by default.
But surprisingly I found in include/asm-x86_64/param.h the following:

#ifndef _ASMx86_64_PARAM_H
#define _ASMx86_64_PARAM_H

#ifdef __KERNEL__
# define HZ 1000 /* Internal kernel timer frequency */
# define USER_HZ 100 /* .. some user interfaces are in "ticks */
#define CLOCKS_PER_SEC (USER_HZ) /* like times() */
#endif

#ifndef HZ
#define HZ 100
#endif

#define EXEC_PAGESIZE 4096

#ifndef NOGROUP
#define NOGROUP (-1)
#endif

#define MAXHOSTNAMELEN 64 /* max length of hostname */

#endif

So i assume the kernel stack size is 4K.
Am i right ? If no how i can make sure the stack size will be 8K to be on the safe side with reiser4?

Lastly the suspend2 do not compile yet on the amd64 platform, again not an issue for me, just for info.

[predatorfreak]

Hi yardbird,

I didn't lost any data as my home partition was still in reiserfs and additionaly i have daily backup of them (36 GB of picture...). I am aventurous but not crasy.
For now i will just move /tmp to reiser4, so if it breaks no issue.
Just one question for my understanding: as far as i know this is no way to chose the kernel stack size in the kernel config (at least on the AMD64 platform).
The parameter is still there but not set and not available in xconfig and not set, so my assumption was a 8k stack size by default.
But surprisingly I found in include/asm-x86_64/param.h the following:

#ifndef _ASMx86_64_PARAM_H
#define _ASMx86_64_PARAM_H

#ifdef __KERNEL__
# define HZ 1000 /* Internal kernel timer frequency */
# define USER_HZ 100 /* .. some user interfaces are in "ticks */
#define CLOCKS_PER_SEC (USER_HZ) /* like times() */
#endif

#ifndef HZ
#define HZ 100
#endif

#define EXEC_PAGESIZE 4096

#ifndef NOGROUP
#define NOGROUP (-1)
#endif

#define MAXHOSTNAMELEN 64 /* max length of hostname */

#endif

So i assume the kernel stack size is 4K.
Am i right ? If no how i can make sure the stack size will be 8K to be on the safe side with reiser4?

Lastly the suspend2 do not compile yet on the amd64 platform, again not an issue for me, just for info.

THREAD_SIZE controls the stack size on x86, since amd64 and x86 are similer (not exact), i would expect that it uses THREAD_SIZE aswell, correct me if i'm wrong.

Edit: After inspecting the varstacks patch and looking at asm-x86_64/thread_info.h (which is generally where the stack-size is defined), it seems as though theres no method for changing the stack size at all on AMD64, also, thats the execute page size, which is a different ballpark then stack size .

[steve_d555]

Hi,
I have always wanted to test your kernel for a long time, but couldnt. Do you think, for the sake of ndiswrapper users that maybe you could have a version of morph sources without Software Suspend 2? It really messes ndiswrapper up, and without it I can not have wireless
Maybe they will fix it in a later version, but I can not be sure.

If you don't have the time/patience, that is ok.

Thank you,
--Steve

[ahorn]

With morph23, i got reiser4 probs an amd64, same with all other kernels. trying now morph24, hope reiser4 runs.
keep your focus on amd64 ! as i said in posts ago, really nice patchset.

edit:

Code: Select all

  CC      mm/swap.o
In file included from mm/swap.c:34:
include/linux/kdb.h:26:21: asm/kdb.h: No such file or directory
make[1]: *** [mm/swap.o] Error 1
make: *** [mm] Error 2

yoshs-knecht root # slocate kdb.h
/usr/src/linux-2.6.10-morph24/include/linux/kdb.h
/usr/src/linux-2.6.10-morph24/include/asm-i386/kdb.h

maybe kdb.h must be in asm-x86_x64/ and in asm/?

ok i did this

Code: Select all

cp /usr/src/linux/include/linux/kdb.h /usr/src/linux/include/asm/

and now it compiles to the end. is there a difference between kdb.h (x86) and kdb.h (x86_64) ?

[yardbird]

Code: Select all

  CC      mm/swap.o
In file included from mm/swap.c:34:
include/linux/kdb.h:26:21: asm/kdb.h: No such file or directory
make[1]: *** [mm/swap.o] Error 1
make: *** [mm] Error 2

yoshs-knecht root # slocate kdb.h
/usr/src/linux-2.6.10-morph24/include/linux/kdb.h
/usr/src/linux-2.6.10-morph24/include/asm-i386/kdb.h

maybe kdb.h must be in asm-x86_x64/ and in asm/?

ok i did this

Code: Select all

cp /usr/src/linux/include/linux/kdb.h /usr/src/linux/include/asm/

and now it compiles to the end. is there a difference between kdb.h (x86) and kdb.h (x86_64) ?

Hi,

this looks like the same problem JBerro reported here. (it should be ok to apply the workaround specified there). It'll be fixed in the next version.