This is an updated patch of the patch that I've posted one year
ago. Shehjar reported the problem and confirmed that the patch fixes
it:
http://lkml.indiana.edu/hypermail/linux/kernel/0811.0/01562.html
Leonardo told me that he hit the same problem and confirmed that the
patch fixes it.
=
From: FUJITA Tomonori <[email protected]>
Subject: [PATCH] ia64: fix SBA IOMMU to handle allocation failure properly
It's possible that SBA IOMMU might fail to find I/O space under heavy
I/Os. SBA IOMMU panics on allocation failure but it shouldn't; drivers
can handle the failure. The majority of other IOMMU drivers don't
panic on allocation failure.
This patch fixes SBA IOMMU path to handle allocation failure properly.
Signed-off-by: FUJITA Tomonori <[email protected]>
---
arch/ia64/hp/common/sba_iommu.c | 38 +++++++++++++++++++++++++++++---------
1 files changed, 29 insertions(+), 9 deletions(-)
diff --git a/arch/ia64/hp/common/sba_iommu.c b/arch/ia64/hp/common/sba_iommu.c
index 674a837..01ae69b 100644
--- a/arch/ia64/hp/common/sba_iommu.c
+++ b/arch/ia64/hp/common/sba_iommu.c
@@ -677,12 +677,19 @@ sba_alloc_range(struct ioc *ioc, struct device *dev, size_t size)
spin_unlock_irqrestore(&ioc->saved_lock, flags);
pide = sba_search_bitmap(ioc, dev, pages_needed, 0);
- if (unlikely(pide >= (ioc->res_size << 3)))
- panic(__FILE__ ": I/O MMU @ %p is out of mapping resources\n",
- ioc->ioc_hpa);
+ if (unlikely(pide >= (ioc->res_size << 3))) {
+ printk(KERN_WARNING "%s: I/O MMU @ %p is"
+ "out of mapping resources, %u %u %lx\n",
+ __func__, ioc->ioc_hpa, ioc->res_size,
+ pages_needed, dma_get_seg_boundary(dev));
+ return -1;
+ }
#else
- panic(__FILE__ ": I/O MMU @ %p is out of mapping resources\n",
- ioc->ioc_hpa);
+ printk(KERN_WARNING "%s: I/O MMU @ %p is"
+ "out of mapping resources, %u %u %lx\n",
+ __func__, ioc->ioc_hpa, ioc->res_size,
+ pages_needed, dma_get_seg_boundary(dev));
+ return -1;
#endif
}
}
@@ -965,6 +972,8 @@ static dma_addr_t sba_map_page(struct device *dev, struct page *page,
#endif
pide = sba_alloc_range(ioc, dev, size);
+ if (pide < 0)
+ return 0;
iovp = (dma_addr_t) pide << iovp_shift;
@@ -1320,6 +1329,7 @@ sba_coalesce_chunks(struct ioc *ioc, struct device *dev,
unsigned long dma_offset, dma_len; /* start/len of DMA stream */
int n_mappings = 0;
unsigned int max_seg_size = dma_get_max_seg_size(dev);
+ int idx;
while (nents > 0) {
unsigned long vaddr = (unsigned long) sba_sg_address(startsg);
@@ -1418,16 +1428,22 @@ sba_coalesce_chunks(struct ioc *ioc, struct device *dev,
vcontig_sg->dma_length = vcontig_len;
dma_len = (dma_len + dma_offset + ~iovp_mask) & iovp_mask;
ASSERT(dma_len <= DMA_CHUNK_SIZE);
- dma_sg->dma_address = (dma_addr_t) (PIDE_FLAG
- | (sba_alloc_range(ioc, dev, dma_len) << iovp_shift)
- | dma_offset);
+ idx = sba_alloc_range(ioc, dev, dma_len);
+ if (idx < 0) {
+ dma_sg->dma_length = 0;
+ return -1;
+ }
+ dma_sg->dma_address = (dma_addr_t)(PIDE_FLAG | (idx << iovp_shift)
+ | dma_offset);
n_mappings++;
}
return n_mappings;
}
-
+static void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
+ int nents, enum dma_data_direction dir,
+ struct dma_attrs *attrs);
/**
* sba_map_sg - map Scatter/Gather list
* @dev: instance of PCI owned by the driver that's asking.
@@ -1493,6 +1509,10 @@ static int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist,
** Access to the virtual address is what forces a two pass algorithm.
*/
coalesced = sba_coalesce_chunks(ioc, dev, sglist, nents);
+ if (coalesced < 0) {
+ sba_unmap_sg_attrs(dev, sglist, nents, dir, attrs);
+ return 0;
+ }
/*
** Program the I/O Pdir
--
1.5.6.5
> This is an updated patch of the patch that I've posted one year
> ago. Shehjar reported the problem and confirmed that the patch fixes
> it:
>
> http://lkml.indiana.edu/hypermail/linux/kernel/0811.0/01562.html
>
> Leonardo told me that he hit the same problem and confirmed that the
> patch fixes it.
>
> =
> From: FUJITA Tomonori <[email protected]>
> Subject: [PATCH] ia64: fix SBA IOMMU to handle allocation failure properly
>
> It's possible that SBA IOMMU might fail to find I/O space under heavy
> I/Os. SBA IOMMU panics on allocation failure but it shouldn't; drivers
> can handle the failure. The majority of other IOMMU drivers don't
> panic on allocation failure.
>
> This patch fixes SBA IOMMU path to handle allocation failure properly.
>
> Signed-off-by: FUJITA Tomonori <[email protected]>
Thanks for refreshing and resubmitting this. I just received feedback
confirming that the patch resolves the problem.
Kind regards,
Leonardo
>>> On 10/18/2009 at 03:07 AM, in message
<[email protected]>, FUJITA Tomonori
<[email protected]> wrote:
> This is an updated patch of the patch that I've posted one year
> ago. Shehjar reported the problem and confirmed that the patch fixes
> it:
>
> http://lkml.indiana.edu/hypermail/linux/kernel/0811.0/01562.html
>
> Leonardo told me that he hit the same problem and confirmed that the
> patch fixes it.
>
> =
> From: FUJITA Tomonori <[email protected]>
> Subject: [PATCH] ia64: fix SBA IOMMU to handle allocation failure properly
>
> It's possible that SBA IOMMU might fail to find I/O space under heavy
> I/Os. SBA IOMMU panics on allocation failure but it shouldn't; drivers
> can handle the failure. The majority of other IOMMU drivers don't
> panic on allocation failure.
>
> This patch fixes SBA IOMMU path to handle allocation failure properly.
Hello,
Please, any chance to have this patch committed to 2.6.33?
Thanks,
Leonardo
>> This patch fixes SBA IOMMU path to handle allocation failure properly.
>
> Please, any chance to have this patch committed to 2.6.33?
Yes. I have it batched to send today.
-Tony