problem: io_remap address exhaustion?

problem: io_remap address exhaustion?

Post by Josh Fryma » Sun, 26 May 2002 02:40:10



hi,

this may be a dumb question, but we're encountering a bit of difficulty with
two PCI cards we have in a system here.

we've got a P4 with 512MB of RAM.  this P4 has two PCI cards, which are
identical, each of which has 256MB of RAM.  we're working on a driver to
talk to the devices and present their entire memory space to the user
via "io_remap()" but it seems to be exhausting the memory space...

eg, if we try to do the remap on the first card, it works fine.  the second
card fails.  if we drop the SDRAM size from 256M to 128M, we can map both
cards fine.  however, we note the starting addresses look like so:

May 24 13:23:33 ilab2 kernel: ixp1200: I21555  CSRs:  phys(0xfe4ff000) virt(0xe98ed000) size(4 KB)
May 24 13:23:33 ilab2 kernel: ixp1200: IXP1200 CSRs:  phys(0xdff00000) virt(0xe992b000) size(1 MB)
May 24 13:23:33 ilab2 kernel: ixp1200: IXP1200 SDRAM: phys(0xc0000000) virt(0xe9a2c000) size(256 MB)
May 24 13:23:33 ilab2 kernel: ixp1200: Hooked & enabled master IRQ(20) handler
May 24 13:23:33 ilab2 kernel: ixp1200: ixp0: Probing/setting up IXP1200 ethernet interface!
May 24 13:23:33 ilab2 kernel: ixp1200: ixp_map_one: Failed to ioremap_nocache() IXP1200's SDRAM memory region!
May 24 13:23:33 ilab2 kernel: ixp1200: Found and initialized 1 IXP1200 board.

is there some place to lower the virtual memory address being used here?  we're
clearly wrapping the address which is causing the failure, but i'm hesitant to
poke in the VM extensively trying to hack a horrible fix for this...

thanks for any pointers (to code or documentation),

josh
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

1. Memory exhaustion in Solaris 2.4

Has anyone noticed the difference in memory requirements in Solaris 2.4
compared to 2.3? Some of the basic processes is using a lot of
memory. When you run /usr/ucb/ps -auxww on a 2.4 and compare it with the
output from a /usr/ucb/ps -auxww on 2.3 you recognize a "slight" increase
in memory use... (about factor 5). This could be a bug in ps but it
seems to be confirmed with vmstat...

Yours /martin

--

    ____                       URL: http://www.cd.chalmers.se/~martin
 __/_|__\__                    Martin Borjesson, Dept 2240, Volvo Data Corp
|  _ |  _  \  <-- A Volvo car. Phone: +46 31 667182 Fax: +46 31 542841

2. 5.8: mutex_lock not supported for g++?

3. memory exhaustion error using 3c503 card on a 386dx/4M RAM:HELP!!!

4. Recommend Mylex SCSI RAID controller?

5. Netscape and swap space exhaustion (was: stable X win mgr)

6. Should know this.... HELP

7. Memory exhaustion exploit/bug?

8. For a small gateway/firewall

9. Netscape and swap space exhaustion

10. Memory exhaustion w/ XF86

11. HTTPD V1.3 Memory Exhaustion

12. ZONE_NORMAL exhaustion (dcache slab)

13. stack exhaustion => "segmentation fault"