On 8/5/19 09:10, Christoph Hellwig wrote:
Seems like we dropped the ball on this..
I still need to test the possible "disable CONFIG_NUMA" workaround. If that works for my single processor rx2800 i2 it could be a good workaround for now, as I assume the older Itanium systems (<= Montvale) won't be affected by such a config change unless they're using those ccNUMA sx1000/sx2000 chipsets.
Did I give you a patch like this (for 5.2 and probably earlier, won't
apply to 5.3-rc) to test before as that is anther idea?
diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c
index 2c2772e9702a..e471158c7c6e 100644
--- a/kernel/dma/direct.c
+++ b/kernel/dma/direct.c
@@ -119,7 +119,8 @@ struct page *__dma_direct_alloc_pages(struct device *dev, size_t size,
}
}
if (!page)
- page = alloc_pages_node(dev_to_node(dev), gfp, page_order);
+ page = alloc_pages_node(local_memory_node(dev_to_node(dev)),
+ gfp, page_order);
if (page && !dma_coherent_ok(dev, page_to_phys(page), size)) {
__free_pages(page, page_order);
It's not the exact same patch as from [1], but the resulting code is identical if I didn't make an error. Cheers, Frank [1]: https://marc.info/?l=linux-ia64&m=156147364328197&w=2
|