LinuxLists.cc - Linux 2.6.25-rc4

2008-03-05 05:03:58

Subject: Linux 2.6.25-rc4

It's a few days late, but I was waiting for some updates for some of the
most annoying regressions until releasing it, so the end result is
hopefully more useful as a result.

In particular, the block layer changes should hopefully have sorted
themselves out, and CD burning etc hopefully works for people again. Same
goes for the the scheduler regressions, and a number of annoying boot-time
problems.

In short, if you had any issues, please do test, and make sure that the
regression list gets updated (whether fixed or not).

The dirstat shows that (as usual) most of the changes are in drivers and
arch (~51% and ~17% respectively), with about half the driver updates
being in network drivers. Full details for those who care:

2.0% Documentation/
3.6% arch/blackfin/
3.6% arch/cris/
2.3% arch/sparc/kernel/
2.8% arch/x86/
17.3% arch/
2.1% drivers/char/
7.0% drivers/dma/
3.0% drivers/firewire/
24.0% drivers/net/
50.6% drivers/
5.6% fs/cifs/
9.4% fs/
3.6% include/
5.2% kernel/
3.9% mm/
2.5% net/

but in general, it's really a fair amount of small changes spread all
over, with most of the changes being quite small (604 commits, most of
them small, with the BNX2X network driver being and the new fsldma driver
the only one that got some bigger changes).

So give it all a good testing,

Linus

2008-03-05 08:14:27

by FUJITA Tomonori

[permalink] [raw]

Subject: Re: Linux 2.6.25-rc4

On Tue, 4 Mar 2008 21:03:44 -0800 (PST)
Linus Torvalds <[email protected]> wrote:

>
> It's a few days late, but I was waiting for some updates for some of the
> most annoying regressions until releasing it, so the end result is
> hopefully more useful as a result.

Ah, the IOMMU patches (alpha and parisc) that I submitted for -mm have
been merged somehow.

The parisc patches were tested but probably the alpha patches not.

Here's a patch for 32bits arch parisc.

=
From: FUJITA Tomonori <[email protected]>
Subject: [PATCH] parisc: fix IOMMU's device boundary overflow bug on 32bits arch

On 32bits boxes, boundary_size becomes zero due to a overflow and we
hit BUG_ON in iommu_is_span_boundary.

Signed-off-by: FUJITA Tomonori <[email protected]>
Cc: Kyle McMartin <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Grant Grundler <[email protected]>
---
drivers/parisc/ccio-dma.c | 4 ++--
drivers/parisc/sba_iommu.c | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/parisc/ccio-dma.c b/drivers/parisc/ccio-dma.c
index 60d338c..62db3c3 100644
--- a/drivers/parisc/ccio-dma.c
+++ b/drivers/parisc/ccio-dma.c
@@ -366,8 +366,8 @@ ccio_alloc_range(struct ioc *ioc, struct device *dev, size_t size)
** ggg sacrifices another 710 to the computer gods.
*/

- boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1, 1 << IOVP_SHIFT);
- boundary_size >>= IOVP_SHIFT;
+ boundary_size = ALIGN((unsigned long long)dma_get_seg_boundary(dev) + 1,
+ 1ULL << IOVP_SHIFT) >> IOVP_SHIFT;

if (pages_needed <= 8) {
/*
diff --git a/drivers/parisc/sba_iommu.c b/drivers/parisc/sba_iommu.c
index e834127..bdbe780 100644
--- a/drivers/parisc/sba_iommu.c
+++ b/drivers/parisc/sba_iommu.c
@@ -341,8 +341,8 @@ sba_search_bitmap(struct ioc *ioc, struct device *dev,
unsigned long shift;
int ret;

- boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1, 1 << IOVP_SHIFT);
- boundary_size >>= IOVP_SHIFT;
+ boundary_size = ALIGN((unsigned long long)dma_get_seg_boundary(dev) + 1,
+ 1ULL << IOVP_SHIFT) >> IOVP_SHIFT;

#if defined(ZX1_SUPPORT)
BUG_ON(ioc->ibase & ~IOVP_MASK);
--
1.5.3.7

2008-03-05 16:47:22

by Grant Grundler

[permalink] [raw]

Subject: Re: Linux 2.6.25-rc4

On Wed, Mar 05, 2008 at 05:09:30PM +0900, FUJITA Tomonori wrote:
> On Tue, 4 Mar 2008 21:03:44 -0800 (PST)
> Linus Torvalds <[email protected]> wrote:
>
> >
> > It's a few days late, but I was waiting for some updates for some of the
> > most annoying regressions until releasing it, so the end result is
> > hopefully more useful as a result.
>
> Ah, the IOMMU patches (alpha and parisc) that I submitted for -mm have
> been merged somehow.
>
> The parisc patches were tested but probably the alpha patches not.
>
> Here's a patch for 32bits arch parisc.
>
> =
> From: FUJITA Tomonori <[email protected]>
> Subject: [PATCH] parisc: fix IOMMU's device boundary overflow bug on 32bits arch
>
> On 32bits boxes, boundary_size becomes zero due to a overflow and we
> hit BUG_ON in iommu_is_span_boundary.
>
> Signed-off-by: FUJITA Tomonori <[email protected]>
> Cc: Kyle McMartin <[email protected]>
> Cc: Matthew Wilcox <[email protected]>
> Cc: Grant Grundler <[email protected]>

Acked-by: Grant Grundler <[email protected]>

thanks,
grant

> ---
> drivers/parisc/ccio-dma.c | 4 ++--
> drivers/parisc/sba_iommu.c | 4 ++--
> 2 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/parisc/ccio-dma.c b/drivers/parisc/ccio-dma.c
> index 60d338c..62db3c3 100644
> --- a/drivers/parisc/ccio-dma.c
> +++ b/drivers/parisc/ccio-dma.c
> @@ -366,8 +366,8 @@ ccio_alloc_range(struct ioc *ioc, struct device *dev, size_t size)
> ** ggg sacrifices another 710 to the computer gods.
> */
>
> - boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1, 1 << IOVP_SHIFT);
> - boundary_size >>= IOVP_SHIFT;
> + boundary_size = ALIGN((unsigned long long)dma_get_seg_boundary(dev) + 1,
> + 1ULL << IOVP_SHIFT) >> IOVP_SHIFT;
>
> if (pages_needed <= 8) {
> /*
> diff --git a/drivers/parisc/sba_iommu.c b/drivers/parisc/sba_iommu.c
> index e834127..bdbe780 100644
> --- a/drivers/parisc/sba_iommu.c
> +++ b/drivers/parisc/sba_iommu.c
> @@ -341,8 +341,8 @@ sba_search_bitmap(struct ioc *ioc, struct device *dev,
> unsigned long shift;
> int ret;
>
> - boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1, 1 << IOVP_SHIFT);
> - boundary_size >>= IOVP_SHIFT;
> + boundary_size = ALIGN((unsigned long long)dma_get_seg_boundary(dev) + 1,
> + 1ULL << IOVP_SHIFT) >> IOVP_SHIFT;
>
> #if defined(ZX1_SUPPORT)
> BUG_ON(ioc->ibase & ~IOVP_MASK);
> --
> 1.5.3.7

2008-03-05 22:30:45

by Jonathan McDowell

[permalink] [raw]

Subject: Re: Linux 2.6.25-rc4

In article
<[email protected]>
(earth.lists.linux-kernel) you wrote:

> It's a few days late, but I was waiting for some updates for some of the
> most annoying regressions until releasing it, so the end result is
> hopefully more useful as a result.

Still crashes in acpi_get_data for me; the patch mentioned
<[email protected]> and tracked at

http://bugzilla.kernel.org/show_bug.cgi?id=10132

still doesn't seem to have hit mainline and still fixes the issue for me.

-----
From: Lin Ming <[email protected]>

Fix a memory overflow bug when copying
NULL internal package element object to external.

Signed-off-by: Lin Ming <[email protected]>
Signed-off-by: Zhang Rui <[email protected]>
---
drivers/acpi/utilities/utobject.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/drivers/acpi/utilities/utobject.c
===================================================================
--- linux-2.6.orig/drivers/acpi/utilities/utobject.c
+++ linux-2.6/drivers/acpi/utilities/utobject.c
@@ -432,7 +432,7 @@ acpi_ut_get_simple_object_size(union acp
* element -- which is legal)
*/
if (!internal_object) {
- *obj_length = 0;
+ *obj_length = sizeof(union acpi_object);
return_ACPI_STATUS(AE_OK);
}

-----

J.

--
/-\ | 101 things you can't have too much
|@/ Debian GNU/Linux Developer | of : 8 - Hard drive space.
\- |

2008-03-06 09:02:03

[permalink] [raw]

Subject: Re: Linux 2.6.25-rc4

> Err.. I have a fairly modern PATA drive (160GB seagate 2.5")
> that gets pissed if we leave DRQ hanging. So it's not just
> something for old/obsolete drives.
>
> Or maybe it's the also-modern-ish ICH chipset that gets stuck.

ICH's with AHCI support seem to get peeved.

Alan