Thanks again. Sorry to be a pest but...
> The arm changes should be split into clean, logical
> chunks and placed in the ARM patch system.
By arm changes do you mean just the DMA stuff or the OHCI SA1111 changes
as well?
> The 2.4.x dma workaround scribbles memory so it
> should be replaced with my code asap.
Do you mean ?
1) Send the dma.patch (previously attached) to rmk now.
[dma workaround = existing code in 2.4.19-rmk2]
[my code = your patch 1255 as modified by me to patch 2.4.19-rmk2]
OR
2) You have more recent code that deals with DMA better than 1255.
[dma workaround = patch 1255 (more or less)]
[my code = something you have that I need]
> I think your Makefiles changes need to be cleaned up
> and simplified
I assume your greatest distaste is for the use of 'findstring' in
drivers/usb/Makefile (line breaks added to fit mail message):
ifneq ($(CONFIG_USB_EHCI_HCD)\
$(CONFIG_USB_OHCI)\
$(CONFIG_USB_OHCI_SA1111),nnn)
usbcore-objs += hcd.o
ifeq ($(findstring y,$(CONFIG_USB_EHCI_HCD)\
$(CONFIG_USB_OHCI)\
$(CONFIG_USB_OHCI_SA1111)),y)
subdir-y += hcd
endif
ifeq ($(findstring m,$(CONFIG_USB_EHCI_HCD)\
$(CONFIG_USB_OHCI)\
$(CONFIG_USB_OHCI_SA1111)),m)
subdir-m += hcd
endif
endif
I actually spent some time trying to find a clean way to do this. I
found findstring in arch/arm/Makefile which is why I felt free to use
it. (You should have seen my earlier mess.) If you really think rmk
will reject it, what do you suggest?
Or have I completely missed the point?
> Those changes alone might make usb-ohci behave.
> But I'm not convinced of that.
Care to elaborate?
> As far as how to proceed with some version of ohci-hcd for 2.4.x,
> that issue should be discussed on the usb mailing list. rmk and
> David Brownell undoubtedly will have opinions about how to go
> about this.
David Brownell <[EMAIL PROTECTED]> says he'd been thinking of applying
the same kind of updates to 'ohci-hcd' that have been applied to
'ehci-hcd' and that I should try 2.5.40 if I want to merge "ohci-hcd"
into 2.4.20.
I've only been looking at 2.5.x's that have been rmk'd so my patch is
for grabbing USB OHCI HCD from 2.5.30 and putting it in 2.4.19. Given
David's response, perhaps I should postpone sharing the usb patch with
the community.
-----Original Message-----
From: Christopher Hoover [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, October 01, 2002 3:33 PM
To: Lackey, Jeff
Cc: Russell King (Russell King); [EMAIL PROTECTED];
'David Brownell'
Subject: RE: Back porting 2.5.x USB to 2.4.x
The arm changes should be split into clean, logical chunks and placed in
the ARM patch system. Those changes are good and appropriate for 2.4.x.
The 2.4.x dma workaround scribbles memory so it should be replaced with
my code asap. (btw -- I think your Makefiles changes need to be cleaned
up and simplified ... I can't imagine rmk taking those as-is.)
Those changes alone might make usb-ohci behave. But I'm not convinced
of that.
As far as how to proceed with some version of ohci-hcd for 2.4.x, that
issue should be discussed on the usb mailing list. rmk and David
Brownell undoubtedly will have opinions about how to go about this.
-ch
-----Original Message-----
From: Lackey, Jeff [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, October 01, 2002 12:13 PM
To: Christopher Hoover
Subject: RE: Back porting 2.5.x USB to 2.4.x
Thanks for the feedback. Excellent advice. Back porting the OHCI HCD
and the DMA fix was much easier than swallowing 2.5.x USB whole.
> Is there some reason you are considering a wholesale back port of
> 2.5 usb?
Well, as you've said yourself regarding the status of sa1111 usb: '2.4.x
for all x is broken'.
I could not get USB OHCI SA-1111 to work at all in 2.4.19-rmk2. With
the attached 2.4.19 patches, USB seems to work better than in my
previous 2.4.x (which was based on 2.4.9-ac10-rmk2-np1).
> there's an on-going effort from the usb maintainers to back port
> *appropriate* 2.5 usb bits to 2.4. I'd suggest you discuss your
> requirements with them if there's something in 2.5 that's missing
> in 2.4 that you need. FYI, if you don't work with the usb
> maintainers, you're unlikely to get anyone upstream to accept
> your patches.
Who should I talk to? I would be happy to contribute the attached
patches if anyone would find them useful. (I should point out that I
don't have the hardware to test USB OHCI PCI.)
Jeff Lackey
Applieddata.net
434.244.9504
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
