I know NET_DMA is currently marked broken upstream, but I am looking
at an issue on RHEL7 where a warning message is getting printed due to
dma_pin_iovec_pages() trying to allocate too large a chunk of memory
for the pinned pages. tcp_recvmsg() looks like it will continue on
fine and not use the dma offloading, so would the correct solution
here be to add __GFP_NOWARN to the kmalloc() or add a check like:
if (get_order(x) >= MAX_ORDER)
goto out;
where x is the size passed to kmalloc?
In the meantime I will look into why NET_DMA is still enabled for us
(kabi perhaps), and hasn't been disabled in some way.
Also is there an archive of the dmaengine mailing list somewhere? It
wasn't listed on the vger.kernel.org page.
Regards,
Jerry
--
To unsubscribe from this list: send the line "unsubscribe dmaengine" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
