Bug 431 - dmaSyncWait error
: dmaSyncWait error
Status: VERIFIED FIXED
: IPW2200
__UNSPECIFIED__
: 0.15
: __UNSPECIFIED__ All
: P2 normal
Assigned To:
:
:
:
:
:
  Show dependency treegraph
 
Reported: 2004-11-29 18:24 by
Modified: 2005-12-05 23:34 (History)


Attachments
compiler output for building version 16 (3.60 KB, text/plain)
2004-12-08 04:54, Allen R. Lorenz
Details


Note

You need to log in before you can comment on or make changes to this bug.


Description From 2004-11-29 18:24:09
Firmware load fails with a dmaWaitSync Failed

Nov 29 18:00:30 Orac kernel: ipw2200: Intel(R) PRO/Wireless 2200/2915 Network
Driver, 0.15
Nov 29 18:00:30 Orac kernel: ipw2200: Copyright(c) 2003-2004 Intel Corporation
Nov 29 18:00:30 Orac kernel: PCI: Enabling device 0000:00:05.0 (0000 -> 0002)
Nov 29 18:00:30 Orac kernel: ACPI: PCI interrupt 0000:00:05.0[A] -> GSI 16
(level, low) -> IRQ 169
Nov 29 18:00:30 Orac kernel: ipw2200: Detected Intel PRO/Wireless 2200BG Network
Connection
Nov 29 18:00:30 Orac kernel: ipw2200: dmaWaitSync Failed
Nov 29 18:00:30 Orac kernel: ipw2200: Unable to load boot firmware
Nov 29 18:00:30 Orac kernel: ipw2200: Unable to load firmware: 0xFFFFFFFF
Nov 29 18:00:30 Orac kernel: ipw2200: failed to register network device
Nov 29 18:00:30 Orac kernel: ipw2200: probe of 0000:00:05.0 failed with error -5

have tested version (12-15) all with same problem and have tried several
combindation of activation of modules and network all give the same error.

Clevo d470k 1 gig ram 
athlong 64 3200 
FC3 (x86_64) all patches ( up2date)
udev-039-10.FC3.2
------- Comment #1 From 2004-12-07 11:21:58 -------
Have you tried the patches referred to in this thread 
http://sourceforge.net/mailarchive/message.php?msg_id=9402251 ?

Would be great if someone in the community provides a patch for this. At this 
point we're not supporting 64-bit, but would be happy to take in a tested 
patch provided it doesn't break 32-bit.
------- Comment #2 From 2004-12-07 12:17:13 -------
Sounds like that might be it....arrggh... Thought I had grepped all the mail 
archive for everything that made sense.  I will try this tonight and email 
status tomorrow... Thanks.
------- Comment #3 From 2004-12-08 04:54:44 -------
Created an attachment (id=148) [details]
compiler output for building version 16

The changes in the command_block structure had already been made.  I believe
you are on the right track as the issue is probably related to 64 v 32 bit. I
have included the compiler output for your further suggestion. 
Compiler version is gcc-3.4.2 here is the -v output if it helps.

Reading specs from /usr/lib/gcc/x86_64-redhat-linux/3.4.2/specs
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man
--infodir=/usr/share/info --enable-shared --enable-threads=posix
--disable-checking --with-system-zlib --enable-__cxa_atexit
--disable-libunwind-exceptions --enable-languages=c,c++,objc,java,f77
--enable-java-awt=gtk --host=x86_64-redhat-linux
Thread model: posix

also uname -a give
Linux Orac 2.6.9-1.681_FC3 #1 Thu Nov 18 15:13:22 EST 2004 x86_64 x86_64 x86_64
GNU/Linux

Looking for more suggestion.  (I tried changing a few types from u32 -> void *
but got more warning/but no change in the error message
------- Comment #4 From 2004-12-08 07:20:35 -------
shouldn't the third argument to pci_alloc_consistant be of type dma_addr_t 
which is u64 ( right ? ) so shouldn't variable shared_phys ( u32) be changed to 
void * ....
------- Comment #5 From 2004-12-10 07:52:40 -------
your right the patch worked, how come this hasn't been integrated into the 
driver... ( x86_64 ) machines (laptops), I would think are at least 5% or more 
of the market which use the ipw2200
------- Comment #6 From 2004-12-10 07:57:14 -------
Re-opening report so that it stays on our radar, and to possibly get the patch 
incoporated into the driver.
------- Comment #7 From 2005-01-13 11:39:07 -------
<logics> there is a patch on the ML i think that is supposed to fix it w/out 
impact to 32-bit, and w/out the removal of the inline
<logics> its basically changing the code to use correct casts of pointers, and 
add a DMA mask to 32-bit land when we allocate pci memory
<logics> which we should have in the driver regardless of if there were a 
64bit issue
<logics> i'm still the owner
<logics> comment #4 is 
<logics> fixed in 0.20 -- you'll see the same problem if you have > 4G of 
memory (or highmem support compiled in)
<logics> dma_addr_t gets set to 64 bits in that case, and we're casting it to 
32 bits
<logics> there are other changes that are not incorporated yet, but i hope to 
get into the driver today
<logics> so set as assigned to me
------- Comment #8 From 2005-01-15 13:05:47 -------
Fixed in 0.20 (known 64-bit patches applied to correct type casting errors,
etc.)
------- Comment #9 From 2005-01-19 23:08:51 -------
Submitter has reported to me that build 0.21 works out of the box for his 64-
bit laptop. Marking as verified.
------- Comment #10 From 2005-12-05 23:34:50 -------
Same problem on my ThinkPad R50e laptop with 1.0.8 when running on batteries.