All of lore.kernel.org
 help / color / mirror / Atom feed
* Problem with remap_page_range on IA32 with more than 4GB RAM
@ 2004-09-23 23:22 John Fusco
  2004-09-23 23:27 ` William Lee Irwin III
                   ` (2 more replies)
  0 siblings, 3 replies; 10+ messages in thread
From: John Fusco @ 2004-09-23 23:22 UTC (permalink / raw)
  To: linux-mm

I have a problem and I would like some comments on how to fix it.

I have a custom PCI-X device installed in an IA32 system.  The device 
expects to see a flat contiguous address space on the host, from which 
it reads and sends its data.  The technique I used is right out of the 
O'Reilly Device Drivers book, which is to hide memory from the kernel 
with the 'mem=YYY' boot parameter.  I then provide a mmap method to map 
the contiguous (hidden) memory into user space via a call to 
'remap_page_range'.

Everything worked great until we decided that we needed to install 6GB 
in this system.  The problem is that remap_page_range() uses an unsigned 
long as the parameter for a physical address.  On IA32, an unsigned long 
is 32-bits, but the IA32 is capable of addressing well over 4GB of RAM.  
So physical addresses on IA32 must be larger than 32 bits.

I chose to work around this by patching the kernel.  I changed the 
unsigned long parameters used for physical address in mm/memory.c to 
'dma64_addr_t'.  This seems to work and I don't see any holes in the 
approach, but I would appreciate any comments (or better solutions).

I can post the patch here if anyone would like to see it.  It seems that 
Linux could use a unique typedef for a physical address.  Right now I 
think dma64_addr_t fits the bill.

Thanks,
John
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

^ permalink raw reply	[flat|nested] 10+ messages in thread

end of thread, other threads:[~2004-09-24  3:39 UTC | newest]

Thread overview: 10+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-09-23 23:22 Problem with remap_page_range on IA32 with more than 4GB RAM John Fusco
2004-09-23 23:27 ` William Lee Irwin III
2004-09-24  2:17 ` [vm 0/4] replace remap_page_range() with remap_pfn_range() William Lee Irwin III
2004-09-24  2:19   ` [vm 1/4] convert remap_page_range() to remap_pfn_range() William Lee Irwin III
2004-09-24  2:21     ` [vm 2/4] convert io_remap_page_range() to call remap_pfn_range() William Lee Irwin III
2004-09-24  2:23       ` [vm 3/4] convert direct callers of remap_page_range() William Lee Irwin III
2004-09-24  2:25         ` [vm 4/4] remove remap_page_range() William Lee Irwin III
2004-09-24  2:27         ` [vm 3/4] convert direct callers of remap_page_range() William Lee Irwin III
2004-09-24  3:29   ` [vm 0/4] replace remap_page_range() with remap_pfn_range() William Lee Irwin III
2004-09-24  3:22 ` Is there a node-affinity memory allocation benchmark? Annie

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.