Development of Linux for the HP PA-RISC platform

Kyle McMartin | 18 Oct 10:29

Picon

what's in parisc-2.6.git

git://git.kernel.org/pub/scm/linux/kernel/git/kyle/parisc-2.6.git
http://www.kernel.org/pub/scm/linux/kernel/git/kyle/parisc-2.6.git

Any last minute things for the 2.6.24 merge window? I'm still sitting on
a lot of development work I've been doing for the last 6 months. Will be
working on getting that merged for 2.6.25 now that things seem to mostly
work from release to release in Linus' tree.

Hooray, this is the first -rc1 cycle where I haven't had to do a
bisect-from-hell to figure out why all my machines won't boot.

[Note, I've given up on the quilt tree for now, because it ended up
being a bit more hassle than just keeping topic branches around.]

Cheers,
	Kyle

Adrian Bunk (1):
      [PARISC] parisc: "extern inline" -> "static inline"

Christoph Lameter (1):
      [PARISC] Use page allocator instead of slab allocator in pci-dma.c

Jeff Bailey (1):
      [PARISC] Export pdc.h for palo

Kyle McMartin (15):
      [PARISC] Wire up sys_fallocate (and compat_sys_fallocate)
      [PARISC] Clean up asm-parisc/pdc.h
      [PARISC] Update defconfigs

(Continue reading)

John David Anglin | 18 Oct 21:37

Picon

Re: what's in parisc-2.6.git

Hi Kyle,

> [Note, I've given up on the quilt tree for now, because it ended up
> being a bit more hassle than just keeping topic branches around.]

Have you also given up on your patch tree?  Seems to have disappeared.

Yesterday, I applied your latest patch set to 2.6.23.1 and built
an SMP kernel.  I didn't test it much but there were weird errors
(e.g., postfix failed to start because it thought user postfix
and user nobody, and apt-get reported missing groups such as
Debian-exim).  This is all bogus and the errors don't occur with
2.6.22.9 + your getdents compat patch.  I still have this kernel
around.

Dave
--

-- 
J. David Anglin                                  dave.anglin <at> nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Kyle McMartin | 18 Oct 21:41

Picon

Re: what's in parisc-2.6.git

On Thu, Oct 18, 2007 at 03:37:39PM -0400, John David Anglin wrote:
> Hi Kyle,
> 
> > [Note, I've given up on the quilt tree for now, because it ended up
> > being a bit more hassle than just keeping topic branches around.]
> 
> Have you also given up on your patch tree?  Seems to have disappeared.
> 

Just temporarily. The workflow wasn't really working for me, and wasn't
motivating me to generate diffs. I'll be putting more -pa testkernels
up.

Now that I have some time again, I'll be fixing the autobuilder as well.

> Yesterday, I applied your latest patch set to 2.6.23.1 and built
> an SMP kernel.  I didn't test it much but there were weird errors
> (e.g., postfix failed to start because it thought user postfix
> and user nobody, and apt-get reported missing groups such as
> Debian-exim).  This is all bogus and the errors don't occur with
> 2.6.22.9 + your getdents compat patch.  I still have this kernel
> around.
> 

Er, really? This is bizarre, I'm running tip on all my machines without
problem (granted, in 64-bit mode.)

Can you upload the failing kernel somewhere, and I'll try to reproduce
it?

(Continue reading)

John David Anglin | 20 Oct 04:14

Picon

Re: what's in parisc-2.6.git

> Just temporarily. The workflow wasn't really working for me, and wasn't
> motivating me to generate diffs. I'll be putting more -pa testkernels
> up.

Ok, I saw you made a lot of updates to parisc-2.6, so I cloned it
this evening.  Noticed a small warning/bug from asm-parisc/rtc.h.

The call to get_rtc_time in get_rtc_ss doesn't check the return
value from get_rtc_time for RTC_BATT_BAD.  As a result, h may
be used unininitialized.  I guess this only happens if the RTC
battery is bad...

Dave
--

-- 
J. David Anglin                                  dave.anglin <at> nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

John David Anglin | 20 Oct 19:24

Picon

Re: what's in parisc-2.6.git

> Ok, I saw you made a lot of updates to parisc-2.6, so I cloned it
> this evening.

This kernel build also has the user/group problem:

...
Starting internet superserver: inetd.
Starting Postfix Mail Transport Agent: postfixpostfix: fatal: file /etc/postfix/main.cf: parameter
default_privs: unknown user name value: nobody
 failed!
...

This is without udev installed.  With udev, the problems are much
more serious:

...
Mount failed for selinuxfs on /selinux:  No such file or directory
INIT: version 2.86 booting
udevd[298]: init_udevd_socket: bind failed: Address already in use
udevd[297]: init_udevd_socket: bind failed: Address already in use
udevd[297]: main: error initializing udevd socket
udevd[296]: main: error disabling OOM: No such file or directory
udevd[299]: init_udevd_socket: bind failed: Address already in use
udevd[299]: main: error initializing udevd socket
* .udev/ already exists on the static /dev!
Starting the hotplug events dispatcher: udevdudevd[379]: init_udevd_socket: bind failed: Address
already in use
error initializing udevd socket
udevd[379]: main: error initializing udevd socket
 failed!

(Continue reading)

John David Anglin | 20 Oct 21:17

Picon

Re: what's in parisc-2.6.git

> Possibly, this caused by setting
> 
>   CONFIG_UEVENT_HELPER_PATH="/sbin/udevd"
> 
> "/sbin/hotplug" doesn't exist and there's no help for this option,
> so it's not obvious what to select.

Changing this back to the default "/sbin/hotplug" avoids this udev
problem, however the kernel HPMC'd at the end of the boot.  Aside
from the hpmc log, there's not much info on what happened.

Back to 2.6.22.9,
Dave
--

-- 
J. David Anglin                                  dave.anglin <at> nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

John David Anglin | 7 Nov 04:32

Picon

Re: what's in parisc-2.6.git

> Changing this back to the default "/sbin/hotplug" avoids this udev
> problem, however the kernel HPMC'd at the end of the boot.  Aside
> from the hpmc log, there's not much info on what happened.
> 
> Back to 2.6.22.9,

I was almost about to say the version as 2007-10-22 resolved the
issues that I have been seeing with 2.6.23.  This build booted ok
but died after three days with a soft lockup:

BUG: soft lockup - CPU#1 stuck for 11s! [udevd:382]

     YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00001000000001000000001100001111 Not tainted
r00-03  000000000804030f 00000000405c6b20 000000004011dde4 0000000000000000
r04-07  00000000405bd320 000000004052ff00 000000004052ff00 00000000405f7008
r08-11  0000000000000001 00000000404a92c0 00000000404b3be0 0000000000000001
r12-15  0000000000000000 00000000fb111bc0 00000000453be3c0 0000000000000000
r16-19  00000000459ecec0 00000000475909c0 00000000453439c0 000000004052ff00
r20-23  0000000000000038 00000000404b4330 0000000000000000 0000000000000001
r24-27  0000000000000000 0000000000000008 00000000404b4330 00000000405bd320
r28-31  0000000000000001 000000004487c750 000000004487c700 00000001032c2993
sr00-03  000000000004a000 0000000000000000 0000000000000000 000000000004a000
sr04-07  0000000000000000 0000000000000000 0000000000000000 0000000000000000

IASQ: 0000000000000000 0000000000000000 IAOQ: 000000004011de24 000000004011de28
 IIR: 0e6010dc    ISR: 00000000474e400a  IOR: 00000000401131cc
 CPU:        1   CR30: 000000004487c000 CR31: fffffff0f0e098e0
ORIG_R28: 0000000000000101
IAOQ[0]: smp_call_function+0x2dc/0x390

(Continue reading)

Carlos O'Donell | 20 Oct 18:44

Re: what's in parisc-2.6.git

On 10/19/07, John David Anglin <dave <at> hiauly1.hia.nrc.ca> wrote:
> > Just temporarily. The workflow wasn't really working for me, and wasn't
> > motivating me to generate diffs. I'll be putting more -pa testkernels
> > up.
>
> Ok, I saw you made a lot of updates to parisc-2.6, so I cloned it
> this evening.  Noticed a small warning/bug from asm-parisc/rtc.h.
>
> The call to get_rtc_time in get_rtc_ss doesn't check the return
> value from get_rtc_time for RTC_BATT_BAD.  As a result, h may
> be used unininitialized.  I guess this only happens if the RTC
> battery is bad...

GCC head is really great at correctly noticing uninitialized uses of
variables. I'm tempted to compile the kernel with head just to fix
these. I just fixed 2 such cases in gdb.

Cheers,
Carlos.

Matthew Wilcox | 20 Oct 19:43

Re: what's in parisc-2.6.git

On Sat, Oct 20, 2007 at 12:44:32PM -0400, Carlos O'Donell wrote:
> GCC head is really great at correctly noticing uninitialized uses of
> variables. I'm tempted to compile the kernel with head just to fix
> these. I just fixed 2 such cases in gdb.

I find it detects too many false positives.  For example ...

int a = 0, b;

for (...) {
	if (c) {
		a = c;
		b = f(c);
	}

	...

	if (a) {
		foo->a = a;
		foor->b = b;
	}
}

warns that b may be used uninitialised.

(reconstructed from memory ... I don't have the exact code to hand right
now)

--

-- 
Intel are signing my paycheques ... these opinions are still mine

(Continue reading)

John David Anglin | 18 Oct 21:56

Picon

Re: what's in parisc-2.6.git

> Er, really? This is bizarre, I'm running tip on all my machines without
> problem (granted, in 64-bit mode.)
> 
> Can you upload the failing kernel somewhere, and I'll try to reproduce
> it?

It's in linux-2.6.23.1 in my home directory on gsyprf11.

Dave
--

-- 
J. David Anglin                                  dave.anglin <at> nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Return

Return to gmane.linux.ports.hppa.

Search Archive

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane