[NITRO] ___Deprecated___

Message

NiLuJe · Post by **NiLuJe** » Tue Mar 21, 2006 10:26 pm

Tiger683: Ok, thanks

Glad my / is on my master hdd ^^

sedorox · Post by **sedorox** » Wed Mar 22, 2006 4:02 am

Trying this now... hopefully it won't break too much stuff (lets see... cisco vpnclient (which isn't working on 15-nitro3 for some reason, probably kernel option), vmware (workstation and server console), madwifi, ati's ****** up drivers...). This is gonna be fun

Also... feel special.. I've NEVER emerged a kernel before.... always done tar, I tend not to trust distro made kernels :p but this is different...

Edit/Update:
Compiled and installed... and this time I decided to try a few things (like running frame buffer with ati's drivers... etc..)
and surprizingly, everything worked!! it came up.. nice text finally on a 1680x1050 screen (laptop), ati's compiled (not clean, but then again I don't think it ever has) but accel works... the cisco vpnclient is working again... vmware compiled and started up normally, the only thing left to try is the madwifi crap. If you want I can post exact specs for this box later so you can keep a eye on things, along with the config I used (which may include a lotta extra crap :p )

Thanks Tiger!! Great job as always!

Tiger683 · Post by **Tiger683** » Wed Mar 22, 2006 9:13 am

sedorox wrote: If you want I can post exact specs for this box later so you can keep a eye on things, along with the config I used (which may include a lotta extra crap :p )

Thanks Tiger!! Great job as always!

Thx

Specs of your box would be nice, the config is not necessary as long as we don't have a problem

The only bug still left to fix is apparently the bloody libata bug with slave devices on ata controller....grrr...

T

Phlogiston · Post by **Phlogiston** » Wed Mar 22, 2006 9:41 am

Tiger683 wrote:
sedorox wrote: If you want I can post exact specs for this box later so you can keep a eye on things, along with the config I used (which may include a lotta extra crap :p )

Thanks Tiger!! Great job as always!
Thx

Specs of your box would be nice, the config is not necessary as long as we don't have a problem

The only bug still left to fix is apparently the bloody libata bug with slave devices on ata controller....grrr...

T

Actally it's not the only one. Since 2.6.15.3 or so I have a hdd spindown during suspend. I think it has to do with one of those sata patches... I filed a bug at kernel.org because it also happens with latest vanilla. If you are interested look here .

@sedorox, did you try the svn ebuilds for madwifi?
And by the way I am also a cisco user

jannis · Post by **jannis** » Wed Mar 22, 2006 1:14 pm

Works like a charm here =D

I could use my server as a mirror, writing a PM now

Phlogiston · Post by **Phlogiston** » Wed Mar 22, 2006 1:43 pm

Using Xgl with this kernel: Xgl eats up all memory after some hours of working with it. Tiger do you think this could be kernel related? (swap prefetch?)

Thanks for any hint

Tiger683 · Post by **Tiger683** » Wed Mar 22, 2006 2:13 pm

I rather think it's an Xgl issue.

Phlogiston · Post by **Phlogiston** » Wed Mar 22, 2006 2:13 pm

Hmm I didn't check sata yet, but thats the problem you're talking about:

Code: Select all

ata1: dev 0 configured for UDMA/100
pata_get_dev_handle: ENTER: dev->bus_id='0000:00:1f.2'
pata_get_dev_handle: dev_handle: 0xdbed37e0, parent_handle: 0xdbec7b20
pata_get_dev_handle: for dev=0x1f.2, addr=0x1f0002, parent=0xdbd36800, *handle=0xdbed37e0
do_drive_set_taskfiles: skipping non-SATA drive

My dvd drive is missing... Do I need to change some boot options?

Thanks!

Tiger683 · Post by **Tiger683** » Wed Mar 22, 2006 2:36 pm

Make sure you have

Code: Select all

libata.atapi_enabled=1

In kernel boot options.

Phlogiston · Post by **Phlogiston** » Wed Mar 22, 2006 2:42 pm

Tiger683 wrote:Make sure you have
Code: Select all
libata.atapi_enabled=1
In kernel boot options.

I have. I just double checked...

Tiger683 · Post by **Tiger683** » Wed Mar 22, 2006 7:02 pm

Is it a slave drive, right?

I have doublechecked and the problem is not related to pata_via itself, so should be present in all
pata drivers i guess

Anyways, i'm looking for a fix the whole time, as this is the last major showstopper
to officially release first stable nitro

PS: After that we will switch to seed tarball + update over git , it'll make the whole thing more flexible

T

tagwar · Post by **tagwar** » Wed Mar 22, 2006 8:27 pm

Hi,

just updated my main box. Most things are working, including nvidia-kernel, fritzcapi, vmware-server, spca5xx, zaptel, ivtv (had to use ivtv-0.6.0).

Now for the not working things:

1) qc-usb (quickcam usb drivers) fails with

Code: Select all

  gcc -m32 -Wp,-MD,/var/tmp/portage/qc-usb-0.6.3/work/qc-usb-0.6.3/.qc-vv6410.o.d  -nostdinc -isystem /usr/lib/gcc/i686-pc-linux-gnu/3.4.5/include -D__KERNEL__ -Iinclude  -include include/linux/autoconf.h -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -ffreestanding -Os     -fomit-frame-pointer -pipe -msoft-float -mpreferred-stack-boundary=2 -fno-unit-at-a-time -march=athlon -Iinclude/asm-i386/mach-default -Wdeclaration-after-statement  -DNOKERNEL   -DMODULE -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(qc_vv6410)"  -D"KBUILD_MODNAME=KBUILD_STR(quickcam)" -c -o /var/tmp/portage/qc-usb-0.6.3/work/qc-usb-0.6.3/qc-vv6410.o /var/tmp/portage/qc-usb-0.6.3/work/qc-usb-0.6.3/qc-vv6410.c
/var/tmp/portage/qc-usb-0.6.3/work/qc-usb-0.6.3/qc-driver.c:3031: error: unknown field `owner' specified in initializer
/var/tmp/portage/qc-usb-0.6.3/work/qc-usb-0.6.3/qc-driver.c:3031: warning: initialization from incompatible pointer type
make[2]: *** [/var/tmp/portage/qc-usb-0.6.3/work/qc-usb-0.6.3/qc-driver.o] Error 1

2) lirc fails with

Code: Select all

make -f scripts/Makefile.build obj=/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c
  gcc -m32 -Wp,-MD,/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/.lirc_i2c.o.d  -nostdinc -isystem /usr/lib/gcc/i686-pc-linux-gnu/3.4.5/include -D__KERNEL__ -Iinclude  -include include/linux/autoconf.h -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -ffreestanding -Os     -fomit-frame-pointer -pipe -msoft-float -mpreferred-stack-boundary=2 -fno-unit-at-a-time -march=athlon -Iinclude/asm-i386/mach-default -Wdeclaration-after-statement  -DIRCTL_DEV_MAJOR=61 -DEXPORT_SYMTAB -DHAVE_CONFIG_H -I. -I. -I../.. -I/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/../.. -I/usr/src/linux//include/  -DMODULE -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(lirc_i2c)"  -D"KBUILD_MODNAME=KBUILD_STR(lirc_i2c)" -c -o /var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.o /var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:363: error: unknown field `name' specified in initializer
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:363: warning: initialization makes integer from pointer without a cast
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:364: error: `I2C_DRIVERID_EXP3' undeclared here (not in a function)
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:364: error: initializer element is not constant
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:364: error: (near initialization for `driver.id')
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:365: error: unknown field `flags' specified in initializer
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:365: error: `I2C_DF_NOTIFY' undeclared here (not in a function)
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:365: error: initializer element is not constant
/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.c:365: error: (near initialization for `driver.class')
make[5]: *** [/var/tmp/portage/lirc-0.8.0/work/lirc-0.8.0/drivers/lirc_i2c/lirc_i2c.o] Error 1

Any ideas?

EDIT: A patch for lirc can be found here: http://sourceforge.net/mailarchive/mess ... d=14677174
EDIT2: Fixed qc-usb myself by commenting out line 3031 in qc-driver.c so it looks like this now:

Code: Select all

static struct usb_driver qc_usb_driver = {
        name:           qc_name,
        probe:          qc_usb_probe,
        disconnect:     qc_usb_disconnect,
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)
// <-- THIS IS WHAT I ADDED     owner :       THIS_MODULE,   
#endif
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,0)
        id_table:       qc_device_table,
#endif
};

Module compiles and loads fine now...

Tiger, especially with the fix for qc-usb, could you perhaps take a look at it and tell if it is totaly stupid what I did? THANKS A LOT!!

Greets Tom

Tiger683 · Post by **Tiger683** » Wed Mar 22, 2006 8:41 pm

Ok, as the stuff is getting a little hairy with bugs and release management, i will
probably push another tree out along with improved distribution method.

No -mm patches and a little more conservative patch-inclusion policy will probably help
us more than anything else now.

I will keep you guys uptodate.

T

sedorox · Post by **sedorox** » Wed Mar 22, 2006 11:15 pm

Update:
Seems I had to remove the radeon driver for the framebuffer to be able to switch between X and VTY's... this has been common before, so I think of it as a problem with ATI's drivers, which it probably is. So no more pretty console for me... allwell...

Madwifi drivers work fine (I didn't try the normal portage, I just did the overlay tiger provided)
ATI Drivers - As said before, compiles (not clean, but never does) but I get accel and it seems to work fine.
Cisco VPN Client - Compiles and runs, connects, but then I can't get out, I'm thinking this is a kernel config on my side (Phlogiston, could I see your network options part of the kernel? if so, PM as to know hold up the thread)
VMWare Workstation - Works fine, (compiled, looked clean :p) used it earlier

This is about the extent of stuff that I have 'third-party' that gets loaded into the kernel.

Specs:
Sager NP8790 Laptop
17" 1680x1050 on a Radeon 9700 Mobility
Intel P4 3.2gig HT, Exteme Edition
1gig ram
2x 60gig drives
Atheros AR5212 Mini-PCI Wifi card
Realtek 8169 gigabit ethernet
3com 3cCFE575BT PCMCIA NIC
Misc USB hardware (printer, dvd burner, etc..)

uname -a: Linux Neptune 2.6.16-nitro_git3 #3 SMP PREEMPT Wed Mar 22 17:11:07 EST 2006 i686 Intel(R) Pentium(R) 4 CPU 3.20GHz GNU/Linux

Tiger683 · Post by **Tiger683** » Thu Mar 23, 2006 8:27 am

tagwar wrote:
Tiger, especially with the fix for qc-usb, could you perhaps take a look at it and tell if it is totaly stupid what I did? THANKS A LOT!!

It's perfectly fine

eNTi · Post by **eNTi** » Thu Mar 23, 2006 9:42 am

has the memory management changed from 2.6.15? because this machine is up for about 1h now and after a little webbrowsing, 62% of my 1GB ram is already full (138MB cached, 183MB buffers).

also my sata controller (ata/0) seems to work a little too much. during downloading it's cpu utilization goes from 0.5% to 5% and somtimes up to 33%.

Tiger683 · Post by **Tiger683** » Thu Mar 23, 2006 9:46 am

This is probably due to adaptive-readahead added in ck. It preloads files into memory when it can be estimated that
they'll be used in a moment

RedshiftBham · Post by **RedshiftBham** » Thu Mar 23, 2006 11:16 am

Hi Tiger.

First of all - an excellent kernel - keep up the good work.

Tiger683 wrote: I have doublechecked and the problem is not related to pata_via itself, so should be present in all
pata drivers i guess

Just to confirm I am having the pata issue on two nForce (a 2 and a 3) boards, i've had to revert back to 2.6.16-git3 since my / is on my slave hdd.

smlgbl · Post by **smlgbl** » Thu Mar 23, 2006 11:27 am

Followed the guide, emerged everything fine, changed to

Code: Select all

samuel@stradivari ~ $ uname -a
Linux stradivari 2.6.16-nitro_git3 #1 Thu Mar 23 11:48:58 CET 2006 i686 AMD Athlon(tm) XP 1600+ AuthenticAMD GNU/Linux

within 1 hour!

My compliments!!!

Tiger683 · Post by **Tiger683** » Thu Mar 23, 2006 1:08 pm

Next commit will be on a reworked tree,
also there will be new overlay and seed tarball.

Stuff will get much cleaner, swsusp got updated to latest development version.
PATA slave issue should be gone too, not a single -mm patch,
probably also custom dsdt support.

Stay tuned, i think i will post it in a new topic to keep things cleaner.

T

smlgbl · Post by **smlgbl** » Thu Mar 23, 2006 3:19 pm

Ehm, just a minor, unimportant thing:

Just after grub, when it boots the kernel, it says: "Can't find font (26pt) from NULL" or something similar.
Doesn't really matter, because it boots just fine after that.

But anyway: Anybody know what that is about?

brot · Post by **brot** » Thu Mar 23, 2006 10:24 pm

Tiger683 wrote:This is probably due to adaptive-readahead added in ck. It preloads files into memory when it can be estimated that
they'll be used in a moment

Since when is adaptive-readahead in your kernel, aaaand does it need to be activated ? I cant remember to see such an option.
And is there anything in /proc that says how effective readahead is ?
Con doesnt really says much about it on his hp

Thanks for your work Tiger. Maybe in the summer vacations i'll visit you and invite you to a free beer

iphitus · Post by **iphitus** » Fri Mar 24, 2006 1:36 am

Phlogiston wrote:Hmm I didn't check sata yet, but thats the problem you're talking about:
Code: Select all
ata1: dev 0 configured for UDMA/100
pata_get_dev_handle: ENTER: dev->bus_id='0000:00:1f.2'
pata_get_dev_handle: dev_handle: 0xdbed37e0, parent_handle: 0xdbec7b20
pata_get_dev_handle: for dev=0x1f.2, addr=0x1f0002, parent=0xdbd36800, *handle=0xdbed37e0
do_drive_set_taskfiles: skipping non-SATA drive
My dvd drive is missing... Do I need to change some boot options?

Thanks!

For some reason the llibata.atapi_enabled=1 option doesnt work for some people. Theres a patch which forces it on, which seems to work better:
http://cvs.archlinux.org/cgi-bin/viewcv ... text/plain

iphitus

Phlogiston · Post by **Phlogiston** » Fri Mar 24, 2006 10:12 am

iphitus wrote:
For some reason the llibata.atapi_enabled=1 option doesnt work for some people. Theres a patch which forces it on, which seems to work better:
http://cvs.archlinux.org/cgi-bin/viewcv ... text/plain

iphitus

Ok I changed that integer in libata-core.c and I think I come closer to the solution:

Code: Select all

libata version 1.20 loaded.
ahci 0000:00:1f.2: version 1.2
ACPI (acpi_bus-0199): Device is not power manageable [20060210]
ahci: probe of 0000:00:1f.2 failed with error -12
ata_piix 0000:00:1f.2: version 1.05-ac1
ata_pci_init_one: pci_dev class+intf: 0x10180
ata_pci_init_one: NO_LEGACY == 0
PCI: Setting latency timer of device 0000:00:1f.2 to 64
ata1: SATA max UDMA/133 cmd 0x1F0 ctl 0x3F6 bmdma 0x18C0 irq 14
ata1: dev 0 cfg 00:045a 49:2b00 82:346b 83:5b29 84:6003 85:3469 86:9a09 87:6003 88:203f 93:600b
ata1: dev 0 ATA-6, max UDMA/100, 117210240 sectors: LBA
ata1(0): applying bridge limits
ata_acpi_push_id: skipping for PATA mode
ata1: dev 0 configured for UDMA/100
pata_get_dev_handle: ENTER: dev->bus_id='0000:00:1f.2'
pata_get_dev_handle: dev_handle: 0xdbed37e0, parent_handle: 0xdbec7b20
pata_get_dev_handle: for dev=0x1f.2, addr=0x1f0002, parent=0xdbd36800, *handle=0xdbed37e0
do_drive_get_GTF:   drive w/ adr=0: v: 0xdbed3c80
do_drive_set_taskfiles: skipping non-SATA drive
scsi0 : ata_piix
  Vendor: ATA       Model: FUJITSU MHV2060A  Rev: 0084
  Type:   Direct-Access                      ANSI SCSI revision: 05
pata_get_dev_handle: ENTER: dev->bus_id='0000:00:1f.2'
pata_get_dev_handle: dev_handle: 0xdbed37e0, parent_handle: 0xdbec7b20
pata_get_dev_handle: for dev=0x1f.2, addr=0x1f0002, parent=0xdbd36800, *handle=0xdbed37e0
ata2: SATA max UDMA/133 cmd 0x170 ctl 0x376 bmdma 0x18C8 irq 15
ata2: SATA port has no device.
scsi1 : ata_piix
pata_get_dev_handle: ENTER: dev->bus_id='0000:00:1f.2'
pata_get_dev_handle: dev_handle: 0xdbed37e0, parent_handle: 0xdbec7b20
pata_get_dev_handle: for dev=0x1f.2, addr=0x1f0002, parent=0xdbd36800, *handle=0xdbed37e0
ata2: SATA port has no device.
SCSI device sda: 117210240 512-byte hdwr sectors (60012 MB)
sda: Write Protect is off
sda: Mode Sense: 00 3a 00 00
SCSI device sda: drive cache: write back
SCSI device sda: 117210240 512-byte hdwr sectors (60012 MB)
sda: Write Protect is off
sda: Mode Sense: 00 3a 00 00
SCSI device sda: drive cache: write back
 sda: sda1 sda2 sda3 sda4 < sda5 sda6 sda7 sda8 sda9 sda10 sda11 sda12 >
sd 0:0:0:0: Attached scsi disk sda
sd 0:0:0:0: Attached scsi generic sg0 type 0

For me interesting lines seem to be: ata2: SATA port has no device. and ata2: SATA port has no device. ? What does this mean? Id the sd driver not ready to use latest libata implementation? Or whats going wrong here?

Thank you so much for your help!

Tiger683 · Post by **Tiger683** » Fri Mar 24, 2006 11:53 am

Ok so i forgot this patch.

Anyways, the problem is still there, slaves keep failing to get detected.

For future: i have some major changes in the eclass, now testing:
1) seed tarball, the download is with resume support and versioning support (i.e. seed-1, seed-2 etc.)

2) useragent handling for hosts that restrict wget downloads (cloaking as IE, Opera etc.)

3) -git4 is done, it will be released in the new release system.

And now more comment on the middle-range future:

Nitro is coming to an end, and it should have happened long ago.
As soon as the release system is ready to go and robust enough to handle just about
anything a patchset needs, nitro will get discontinued.

Reason?
Me and iphitus are of the same attitude that making one patchset instead of
competing between them is a better idea. So, we will keep maintaining a new patchset
together

(No hints on name yet

)

I rely on iphitus in matter of keeping things stable and thus i think this
change will be benefical for everyone, while keeping the patchset featurerich

cheers

T