Roland Dreier [EMAIL PROTECTED] wrote on 10.11.2008 21:36:23:
A previous fix introduced a regression where port activation events
were
dropped unconditionally if port autodetection was not enabled. Fixed.
Is this a fix to IB/ehca: Remove reference to special QP in case of
port
Linus,
Please pull from the 'merge' branch of
git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc.git merge
It's mostly defconfig updates, but there are also fixes for a few
regressions, oopses, serious bugs, and documentation errors on
powerpc.
Thanks,
Paul.
Sorry for the slow reply, but my shell account is broken and I had to
post from home.
On Nov 10, 2008, at 10:24 AM, Ilya Yanok wrote:
This patch adds support for 256K pages on PPC 44x along with
some hacks needed for this.
This description is insufficient, it describes neither the hacks nor
On Tue, 2008-11-11 at 08:19 -0500, Josh Boyer wrote:
On Fri, Oct 31, 2008 at 06:23:28PM -0500, Hollis Blanchard wrote:
On Wed, Oct 22, 2008 at 9:28 AM, Christian Ehrhardt
[EMAIL PROTECTED] wrote:
Hi Ilya,
I just tried your patch on my 440 board because it would help us in our
environment.
We take a vanilla kernel with rt11 patches from Ingo Mollnar.
Here are the section of my boot messages
ata: MPC52xx IDE/ATA libata driver
scsi0 : mpc52xx_ata
ata1: PATA max UDMA/33 ata_regs 0xf0003a00 irq 135
ata1.00: ATA-0: ELITE PRO CF CARD 1GB, 20070131, max MWDMA2
ata1.00: 2031120 sectors,
Before freeing an endpoint's muram memory, we should stop all activity
of the endpoint, otherwise the QE UDC controller might do nasty things
with the muram memory that isn't belong to that endpoint anymore.
The qe_ep_reset() effectively flushes the hardware fifos, finishes all
late transaction
The call chain is this:
qe_udc_irq() - grabs the udc-lock spinlock
rx_irq()
qe_ep0_rx()
ep0_setup_handle()
setup_received_handle()
ch9getstatus()
qe_ep_queue() - tries to grab the udc-lock again
It seems unsafe to temporarily drop the lock in the ch9getstatus(),
so to fix that bug the
On Nov 5, 2008, at 10:42 PM, Paul Mackerras wrote:
Milton Miller writes:
With the new generic smp call function helpers, I noticed the code in
smp_message_recv was a single function call in many cases. While
getting the message number from the ipi data is easy, we can reduce
the path length
Hmm even with DMA built in my disk (Seagate 5400.8) still configured for
PIO4.
This is probably my fault. Tim, could you respin the patch anyway since
it doesn't apply cleanly to the 2.6.27 kernel? :)
This did work for me once but we were still using a SUSE kernel (from
11.0) to patch it onto
Hi all,
A few more qe_udc fixes...
--
Anton Vorontsov
email: [EMAIL PROTECTED]
irc://irc.freenode.net/bd2
___
Linuxppc-dev mailing list
Linuxppc-dev@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-dev
On Tue, 2008-11-11 at 11:29 -0600, Matt Sealey wrote:
(the other tool we need is mkimage from U-Boot, I noticed Debian uses
this
same command name inside jigdo etc. too, sigh.. also you can't build
the
tools without the rest of U-Boot which is hard to envision doing just
to
get one command
On Tue, 11 Nov 2008 14:10:39 +
David Woodhouse [EMAIL PROTECTED] wrote:
You might do better to open the device once and keep it open, rather
than taking the mutex and opening it again _during_ each call? You're
going to be perturbing the timing by doing that.
If you keep it open no other
On Nov 10, 2008, at 9:22 AM, Mohan Kumar M wrote:
Yes, elf_ret_symbol sets the copy in purgatory. But the following code
in purgatory (to be introduced)
LOADADDR(6,run_at_load)
ld 18,0(6)
cmpd18,1
bne skip
li 7,1
stw 7,92(4)
qe_udc_reg_init() leaves the USB controller enabled before muram memory
initialized. Sometimes the uninitialized muram memory confuses the
controller, and it start sending the busy interrupts.
Fix this by disabling the controller, it will be enabled later by
the gadget driver, at bind time.
With this patch we can compile the qe_lib/usb.c without the UCC
support (that is, without UCC_GETH and/or SERIAL_QE).
Fixes following link error (CONFIG_SMP should be =y to trigger this):
arch/powerpc/sysdev/built-in.o: In function `qe_usb_clock_set':
(.text+0x3cae): undefined reference to
In case of probing errors the driver kfrees the udc_controller, but it
doesn't set the pointer to NULL.
When usb_gadget_register_driver is called, it checks for udc_controller
!= NULL, the check passes and the driver accesses nonexistent memory.
Fix this by setting udc_controller to NULL in case
Tim, Grant,
I have tested the patch on our MPC5200B based board with Kingston Elite Pro
CF-card and it worked fine with DMA2. I have port the patch back to 2.6.24.7
Kernel and I can't detect problems our errors, even with big files (40 mB) over
FEC wich I save to card . We get performance up
On Tuesday 11 November 2008, David Woodhouse wrote:
I believe you were also concerned that some device wouldn't want the
behaviour given by the existing sync_cmos_clock() function and workqueue
stuff in kernel/ntp.c, where we update the clock precisely half-way
through the second?
That's a
On Mon, 2008-11-10 at 16:40 +0100, Alessandro Zummo wrote:
Adds in-kernel hctosys functionality that can
be used by ntp sync code.
This is an RFC and has not been tested, I just want
to check if something similar could solve the problems
of those who want the NTP sync mode.
You might do
On Tue, Nov 11, 2008 at 02:10:39PM +, David Woodhouse wrote:
On Mon, 2008-11-10 at 16:40 +0100, Alessandro Zummo wrote:
Adds in-kernel hctosys functionality that can
be used by ntp sync code.
This is an RFC and has not been tested, I just want
to check if something similar could
On Fri, Oct 31, 2008 at 06:23:28PM -0500, Hollis Blanchard wrote:
On Wed, Oct 22, 2008 at 9:28 AM, Christian Ehrhardt
[EMAIL PROTECTED] wrote:
Hi Ilya,
I just tried your patch on my 440 board because it would help us in our
environment.
Unfortunately I run into a bug on early boot
On Mon, Nov 10, 2008 at 05:24:34PM -0600, Hollis Blanchard wrote:
The current CHIP11 errata truncates the device tree memory node, and subtracts
(hardcoded) 4096 bytes. This breaks kernels with larger PAGE_SIZE, since the
bootmem allocator assumes that total memory is a multiple of PAGE_SIZE.
In exactly the same way that we updated memcpy() with new feature sections
in commit 25d6e2d7c58ddc4a3b614fc5381591c0cfe66556 we do the same thing
here for __copy_tofrom_user(). Once again this is purely a performance
tweak for Cell and Power6 - this has no effect on all the other 64bit
powerpc
Oddly I built 2.6.27.2 with the patch and the Efika waited for a disk to
appear
and didn't get anywhere. If I set libata.dma=0 then it complained about missing
symbols (but not with it enabled!).
This was a clean SUSE kernel from Factory, with only the ATA patch applied with
a few hand-merges
On Tue, Nov 11, 2008 at 9:31 AM, Anton Vorontsov
[EMAIL PROTECTED] wrote:
With this patch we can compile the qe_lib/usb.c without the UCC
support (that is, without UCC_GETH and/or SERIAL_QE).
Fixes following link error (CONFIG_SMP should be =y to trigger this):
For some of what we're doing with SUSE now I've determined we should really
be packaging the device tree compiler (and libfdt) so that we can install and
script it, along with some other stuff from U-Boot (not for this mail).
I was wondering if the recommended way would be to use the dtc found
Freescale QE UDC controllers can't report the port change states,
so the only way to handle disconnects is to process bus reset
interrupts. The bus reset can take some time, that is, few irqs.
Gadgets may print the disconnection events, and this causes few
repetitive messages in the kernel log.
While disabling an endpoint the driver nuking any pending requests,
thus completing them with -ESHUTDOWN status. But the driver doesn't
clear the tx_req, which means that a next TX request (after
ep_enable), might get stalled, since the driver won't queue the new
reqests.
This fixes a bug I'm
On Tue, Nov 11, 2008 at 11:29:57AM -0600, Matt Sealey wrote:
For some of what we're doing with SUSE now I've determined we should really
be packaging the device tree compiler (and libfdt) so that we can install and
script it, along with some other stuff from U-Boot (not for this mail).
I was
On Nov 10, 2008, at 11:00 AM, Jimi Xenidis wrote:
On Nov 10, 2008, at 10:11 AM, Milton Miller wrote:
On 2008-11-07 at 02:31:40, David Gibson wrote:
On Thu, Nov 06, 2008 at 06:55:44PM +1100, Michael Ellerman wrote:
This commit adds an output format, which produces python
code. When run, the
On Wed, 2008-11-12 at 09:39 +1100, David Gibson wrote:
Use the tree from jdl.org. Either the git tree if you really need an
up-to-date snahpshot, or one of the actual release tarballs.
The tree from jdl.com will work better. :-)
jdl
___
Hi,
On Tue, Nov 11, 2008 at 11:29:57AM -0600, Matt Sealey wrote:
(the other tool we need is mkimage from U-Boot, I noticed Debian uses this
same command name inside jigdo etc. too, sigh.. also you can't build the
tools without the rest of U-Boot which is hard to envision doing just to
get
On Tue, Nov 11, 2008 at 07:52:18AM -0500, Josh Boyer wrote:
On Mon, Nov 10, 2008 at 05:24:34PM -0600, Hollis Blanchard wrote:
The current CHIP11 errata truncates the device tree memory node, and
subtracts
(hardcoded) 4096 bytes. This breaks kernels with larger PAGE_SIZE, since the
bootmem
The current CHIP11 errata truncates the device tree memory node, and subtracts
(hardcoded) 4096 bytes. This breaks kernels with larger PAGE_SIZE, since the
bootmem allocator assumes that total memory is a multiple of PAGE_SIZE.
Instead, use a device tree memory reservation to reserve only the 256
On Tue, Nov 11, 2008 at 06:06:46PM -0600, Hollis Blanchard wrote:
The current CHIP11 errata truncates the device tree memory node, and subtracts
(hardcoded) 4096 bytes. This breaks kernels with larger PAGE_SIZE, since the
bootmem allocator assumes that total memory is a multiple of PAGE_SIZE.
A powerpc ppc64_defconfig build of Linus' current tree produces these
warnings:
fs/ext4/balloc.c: In function 'ext4_has_free_blocks':
fs/ext4/balloc.c:617: warning: format '%lld' expects type 'long long int', but
argument 2 has type 's64'
fs/ext4/inode.c: In function 'ext4_print_free_blocks':
Those cores use the 440A type machine check (ie, they have
MCSRR0/MCSRR1). They thus need to call the appropriate fixup
function to hook the right variant of the exception.
Without this, all machine checks become fatal due to loss
of context when entering the exception handler.
Signed-off-by:
There's a problem on some embedded platforms when we re-assign
everything on PCI, such as 44x. The generic code tries to avoid
assigning devices to addresses overlapping the low legacy
addresses such as VGA hard decoded areas using constants that
are unfortunately no good for us, as they don't
Hi,
OK, another go at submitting these 3 memory ordering improvement patches.
I've tested on my G5, but I'd like to especially get confirmation as to
whether the __SUBARCH_HAS_LWSYNC thing works on E500MC.
--
Change 2d1b2027626d5151fff8ef7c06ca8e7876a1a510 removed __SUBARCH_HAS_LWSYNC,
causing
After commit 598056d5af8fef1dbe8f96f5c2b641a528184e5a, rmb() becomes a sync
instruction, which is needed to order cacheable vs noncacheable loads. However
smp_rmb() is #defined to rmb(), and smp_rmb() can be an lwsync.
Restore smp_rmb() performance by using lwsync there. Update comments.
Implement a more optimal mutex fastpath for powerpc, making use of acquire
and release barrier semantics. This takes the mutex lock+unlock benchmark
from 203 to 173 cycles on a G5.
Signed-off-by: Nick Piggin [EMAIL PROTECTED]
---
Index: linux-2.6/arch/powerpc/include/asm/mutex.h
On Tue, 2008-11-11 at 18:06 -0600, Hollis Blanchard wrote:
The current CHIP11 errata truncates the device tree memory node, and subtracts
(hardcoded) 4096 bytes. This breaks kernels with larger PAGE_SIZE, since the
bootmem allocator assumes that total memory is a multiple of PAGE_SIZE.
42 matches
Mail list logo