[PATCH] yenta: add modparam to tune EnE bridges (was Re: ENE CB1410 and Echo Indigo IO incompatibility)

Daniel Ritz daniel.ritz at gmx.ch
Wed Aug 24 13:13:05 EDT 2005


On Tuesday 23 August 2005 22.39, Jon Stutters wrote:
> Daniel Ritz wrote:
> > On Sunday 21 August 2005 21.10, Jon Stutters wrote:
> > 
> >>If I read the source you attached to the previous message correctly then 
> >>you're still only dealing with the TLTEnable bit.  The setting I had to 
> >>adjust to fix the echo audio card was prefetch which is controlled by 
> >>bit C8.
> >>
> > 
> > <start quoting>
> > The prefetch function can be disabled under Linux by changing C8h to 
> > 0200h in the cardbus chipset
> > </end quoting>
> > 
> > by reading this i was thinking its only about that bit because you have
> > the TLTEnable bit set, but the code clears it by default...
> > 
> > i now found some crappy doc from ene. reading it it's probably bit 2/3 at c9h
> > to control prefetch. btw. don't change anything in c8, only change c9, 8bit.
> > 
> > could you give me the content of that register before you change it, please.
> > ( hexdump /proc/bus/pci/02/04.0 )
> > but anyway, try the attached updated auto-tuning patch.
> 
> Output of hexdump before making any changes is below.  The output 
> doesn't change after I use my setpci fix.
> 
> jstutters at nevis> hexdump /proc/bus/pci/02/04.0
> 0000000 1524 1410 0007 0210 0001 0607 a810 0002
> 0000010 0000 9010 00a0 0200 0302 b006 0000 2040
> 0000020 f000 207f 0000 2080 f000 20bf 5000 0000
> 0000030 50fc 0000 5400 0000 54fc 0000 0105 0500
> 0000040 0e11 0860 0001 0000 0000 0000 0000 0000
> 0000050 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 0000080
> 

why it doesn't show the whole config space (256 bytes).
the register is at c9...so could you just to the setpci
without the new value (no =...)?

> Which kernel source are you developing patches again?  The most recent
> patch you sent didn't apply cleanly to vanilla 2.6.12.5.

the patch is against 2.6.13-rc6 or more exaclty git-head. what
do you mean by not applying cleanly? just some offsets or some
failed hunks?

> Cheers,
> Jon Stutters
> 

rgds
-daniel



More information about the linux-pcmcia mailing list