On Mon, 2012-06-04 at 16:43 +1000, Michael Ellerman wrote:
> There is some chance this will result in breakage because the driver
> asks for N - and assumes that is what was allocated - and the device is
> configured for > N.
We can fix that. We can whack the configuration back with N, just know
t
On Mon, 2012-06-04 at 09:15 +1000, Anton Blanchard wrote:
> The pseries firmware currently refuses any non power of two MSI-X
> request. Unfortunately most network drivers end up asking for that
> because they want a power of two for RX queues and one or two extra
> for everything else.
>
> This p
At the moment all queues in a multiqueue adapter will serialise
against the IOMMU table lock. This is proving to be a big issue,
especially with 10Gbit ethernet.
This patch creates 4 pools and tries to spread the load across
them. If the table is under 1GB in size we revert back to the
original b
In preparation for IOMMU pools, push the spinlock into
iommu_range_alloc and __iommu_free.
Signed-off-by: Anton Blanchard
---
Index: linux-build/arch/powerpc/kernel/iommu.c
===
--- linux-build.orig/arch/powerpc/kernel/iommu.c
This patch moves tce_free outside of the lock in iommu_free.
Some performance numbers were obtained with a Chelsio T3 adapter on
two POWER7 boxes, running a 100 session TCP round robin test.
Performance improved 25% with this patch applied.
Signed-off-by: Anton Blanchard
---
Index: linux-buil
We currently hold the IOMMU spinlock around tce_build and tce_flush.
This causes our spinlock hold times to be much higher than required
and can impact multiqueue adapters.
This patch moves tce_build and tce_flush outside of the lock in
iommu_alloc, and tce_flush outside of the lock in iommu_free
tce_buildmulti_pSeriesLP uses a per cpu page to communicate with the
hypervisor. We currently rely on the IOMMU table spinlock but
subsequent patches will be removing that so disable interrupts
around all accesses of tce_page.
Signed-off-by: Anton Blanchard
---
Index: linux-build/arch/powerpc/p
Hi Paul,
> This reverts 68568add2c ("powerpc/time: Remove unnecessary sanity
> check of decrementer expiration"). We do need to check whether we
> have reached the expiration time of the next event, because we
> sometimes get an early decrementer interrupt, most notably when we
> set the decremen
The pseries firmware currently refuses any non power of two MSI-X
request. Unfortunately most network drivers end up asking for that
because they want a power of two for RX queues and one or two extra
for everything else.
This patch rounds up the firmware request to the next power of two
if the q
Benjamin Herrenschmidt wrote on 2012/06/02 23:21:16:
>
> On Sat, 2012-06-02 at 20:29 +0200, Joakim Tjernlund wrote:
> >
> > hmm, where does this go w.r.t the patch? Got the feeling that the
> > best thing is to just turn MSR:DE on and be done with it?
>
> Not unconditionally, we need to have a clo
All SOC device error interrupts are muxed and delivered to the core as a single
MPIC error interrupt. Currently all the device drivers requiring access to
device
errors have to register for the MPIC error interrupt as a shared interrupt.
With this patch we add interrupt demuxing capability in the
We should use the MPIC_LARG_VECTORS flag while intializing the MPIC.
This prevents us from eating in to hardware vector number space (MSIs)
while setting up internal sources.
Signed-off-by: Varun Sethi
---
arch/powerpc/sysdev/mpic.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
d
Previously, these interrupts would be mapped, but the offset
calculation was broken, and only the first group was initialized.
Signed-off-by: Scott Wood
---
arch/powerpc/include/asm/mpic.h |5 +++
arch/powerpc/sysdev/mpic.c | 58 ---
2 files changed
This patchset adds/fixes the following functionality specific to the
FSL MPIC:
1. Fix support for timer group B interrupts. Previously these were
not getting initialized.
2. Use the MPIC_LARGE_VECTORS flag while intializing FSL MPIC.
This prevents us from eating in to hardware vector number
space
On 06/03/2012 02:59 PM, tiejun.chen wrote:
> On 06/03/2012 01:07 PM, Tiejun Chen wrote:
>> We don't do the real store operation for kprobing 'stwu Rx,(y)R1'
>> since this may corrupt the exception frame, now we will do this
>> operation safely in exception return code after migrate current
>> excep
On 06/03/2012 01:07 PM, Tiejun Chen wrote:
> We don't do the real store operation for kprobing 'stwu Rx,(y)R1'
> since this may corrupt the exception frame, now we will do this
> operation safely in exception return code after migrate current
> exception frame below the kprobed function stack.
>
>
16 matches
Mail list logo
