ADI Blackfin porting for kernel-2.6.13
Hi,
I am Luke Yang, an engineer from Analog Devices Inc. We ported
uclinux to our Blackfin cpu. Now we updated our architecture code for kernel-2.6.13. I will send out a patch to this list.
I know kernel-2.6.14 is coming. Will the linux kernel accept our
patch for 2.6.13?
Regards,
Luke Yang
On Sep 20 2005, at 14:33, Luke Yang was caught saying:
Quoted 8 lines
Hi,
I am Luke Yang, an engineer from Analog Devices Inc. We ported
uclinux to our Blackfin cpu. Now we updated our architecture code for kernel-2.6.13. I will send out a patch to this list.I know kernel-2.6.14 is coming. Will the linux kernel accept our
patch for 2.6.13?
Nope. 2.6.13 is now closed to new features as is 2.6.14 (unless it is a really sper special case that Linus feels is important enough to slip in). At this point the best thing to do is to post your patches for review, make the changes that are asked, and be ready to post a patch vs 2.6.14 within the first week after it is released so that it might be be picked up for 2.6.15.
~Deepak
Signature · 3 lines
Deepak Saxena - [email protected] - http://www.plexity.net Even a stopped clock gives the right time twice a day.
Hi all,
This is the ADI Blackfin patch for kernel 2.6.13. I know this
patch doesn' meet the kernel patch submission format, but this patch is only for reviewing, shows the changes we made. And I'll send new patch for kernel 2.6.14 for you to merge into kernel.
This patch mainly includes the arch/bfinnommu architecture files
and some blackfin specific drivers. The only change to the common files is that we change binfmt.c and related header file, added one flat binary format for blackfin. We are considering removing this change in next release.
The patch is too big to put here , url is
http://blackfin.uclinux.org/frs/download.php/570/bfinnonnu-linux-2.6.13.patch
Thanks!
Luke
Quoted 24 lines · Deepak Saxena
On Sep 20 2005, at 14:33, Luke Yang was caught saying:
Quoted 8 lines
Hi,
I am Luke Yang, an engineer from Analog Devices Inc. We ported
uclinux to our Blackfin cpu. Now we updated our architecture code for kernel-2.6.13. I will send out a patch to this list.I know kernel-2.6.14 is coming. Will the linux kernel accept our
patch for 2.6.13?Nope. 2.6.13 is now closed to new features as is 2.6.14 (unless it is a really sper special case that Linus feels is important enough to slip in). At this point the best thing to do is to post your patches for review, make the changes that are asked, and be ready to post a patch vs 2.6.14 within the first week after it is released so that it might be be picked up for 2.6.15.
~Deepak
--
Deepak Saxena - [email protected] - http://www.plexity.netEven a stopped clock gives the right time twice a day.
Quoted 2 lines · Luke Yang
This patch mainly includes the
arch/bfinnommuarchitecture files and some blackfin specific drivers.
Having in mind the pain with arm and m68k with and without MMU, could you structure that thing in a way that it will be able to share the current nommu code with (hopefully coming) code for blackfins which might have a MMU? So it would be something like arch/blackfin.
Robert
Signature · 5 lines
Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
Pengutronix - Linux Solutions for Science and Industry
Handelsregister: Amtsgericht Hildesheim, HRA 2686
Hannoversche Str. 2, 31134 Hildesheim, Germany
Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9Quoted 7 lines · Luke Yang
Hi all,
This is the ADI Blackfin patch for kernel 2.6.13. I know this
patch doesn' meet the kernel patch submission format, but this patch is only for reviewing, shows the changes we made. And I'll send new patch for kernel 2.6.14 for you to merge into kernel.
Quoted 9 lines
This patch mainly includes the
arch/bfinnommuarchitecture files
and some blackfin specific drivers. The only change to the common files is that we changebinfmt.cand related header file, added one flat binary format for blackfin. We are considering removing this change in next release.The patch is too big to put here , url is
http://blackfin.uclinux.org/frs/download.php/570/bfinnonnu-linux-2.6.13.patch
Hi Luke,
Since you ask for review I decided to give you a hand.
I've started to look through your patch and started out by looking through the documentation bits
Documentation/blackfin/00-INDEX
Documentation/blackfin/Filesystems
Documentation/blackfin/cache-lock.txt
Documentation/uml/UserModeLinux-HOWTO.txt
and I have a few suggested changes (see attached patch).
I'll be taking a look at the rest of the patch as well, just wanted to send you the bits I'd already done for the documentation - its a pretty large patch, so I'm going through it in little bits :)
Hope this is useful to you. More feedback on its way.
Signature · 1 line
Jesper Juhl <[email protected]>
Documentation/blackfin/00-INDEX +2 -2
@@ -5,7 +5,7 @@ - HOWTO for blackfin cache locking. cachefeatures.txt - - Supported cache features + - Supported cache features. Filesystems - - Required for mounting the root file systems. + - Requirements for mounting the root file system.
Documentation/blackfin/Filesystems +37 -33
156 lines not on this page · raw message
Documentation/blackfin/cache-lock.txt +8 -8
@@ -1,31 +1,31 @@ -How to Lock your code into the cache in uClinux/blackfin --------------------------------------------------------- +How to lock your code in cache in uClinux/blackfin +-------------------------------------------------- -There are only few steps required to lock your code into the cache. +There are only a few steps required to lock your code into the cache. Currently you can lock the code by Way. Below are the interface provided for locking the cache. + 1. cache_grab_lock(int Ways); This function grab the lock for locking your code into the cache specified by Ways. + 2. cache_lock(int Ways); This function should be called after your critical code has been executed. -Once the critical code exits, the code is now loaded into the cache.This +Once the critical code exits, the code is now loaded into the cache. This function locks the code into the cache. So, the example sequence will be: cache_grab_lock(WAY0_L); /* Grab the lock */ - + critical_code(); /* Execute the code of interest */ - cache_lock(WAY0_L); /* Lock the cache*/ + cache_lock(WAY0_L); /* Lock the cache */ Where WAY0_L signifies WAY0 locking. - -
Documentation/uml/UserModeLinux-HOWTO.txt +266 -277
1,499 lines not on this page · raw message
Quoted 7 lines · Luke Yang
Hi all,
This is the ADI Blackfin patch for kernel 2.6.13. I know this
patch doesn' meet the kernel patch submission format, but this patch is only for reviewing, shows the changes we made. And I'll send new patch for kernel 2.6.14 for you to merge into kernel.
Quoted 9 lines
This patch mainly includes the
arch/bfinnommuarchitecture files
and some blackfin specific drivers. The only change to the common files is that we changebinfmt.cand related header file, added one flat binary format for blackfin. We are considering removing this change in next release.The patch is too big to put here , url is
http://blackfin.uclinux.org/frs/download.php/570/bfinnonnu-linux-2.6.13.patch
Ok, I went through the files in arch/bfinnommu/kernel/ and I have some comments (I don't have the time to look at the rest of it). I think you need to do some work before merging this.
I've attached a patch (compressed due to its size) that contain all the changes I refer to below. It's just one big patch with all the changes I made while I went through the files. It's not split out into logically seperate changes - sorry. I just made different types of changes along the way, and the attached patch is the result.
You need to clean this stuff up with respect to CodingStyle. Currently the files are quite inconsistent in the style(s) they use, and they don't follow the kernels CodingStyle very closely either.
You have a lot of very deeply nested code (especially in dma.c) that is quite hard to follow and also violates the 80column rule. Try to reduce the nesting depth.
I've made most of the code fit in 80 columns, but I've probably missed a few spots.
the *_interrupt functions in dma.c are just begging for a few helper functions to share common code and reduce complexity. I've reordered them a bit to make them more readable, but there's lots more that could be done.
There's *lots* of trailing whitespace all over the place. Please get rid of that prior to merging.
There's *lots* of cases of mixing of spaces and tabs for indentation. Indent with tabs only please, and certainly don't have lines like "<space><tab><space><space><space><space><tab>code"
like you have at the moment.
I've cleaned up all the trailing whitespace, and some of the other space+tab braindamage, but there's probably still something left to do.
There are quite a lot of casts in the code, and some of them are completely pointless. There's no need to cast the return value of kmalloc(), there's no need to cast assignments to/from void pointers, there's no need to cast assignments between types where normal C promotion rules obtain the same result.
Such unnesesary casting is harmful. It defeats what limited typechecking C has, so the compiler can't help and warn when you do something bad, it also lessens the usefullnes of tools such as sparse.
I've cleaned up some of this, but not everything.
You seem to be quite fond of MixingCapsAndLowercaseChars in variable names. Please don't unless you really have to. Stick to purely lowercase for variable names as much as possible.
I've renamed a few vars, just as examples.
I stumbled across some unused variables as well that I removed, but there may be more.
I've cleaned up a lot of other CodingStyle issues as well. Check the attached patch and you'll see.
In dma.c :
DMA_RESULT set_dma_transfer_size(unsigned int channel, char size) DMA_RESULT get_dma_transfer_size(unsigned int channel, unsigned short *size)
Seems inconsistent. Why "char size" for set_dma_transfer_size(), but "unsigned short *size" for get_dma_transfer_size() ??
An if char is what you want to use for size in set_dma_transfer_size(), then shouldn't that be "unsigned char" ?
I think that's all, read the patch (or apply it and look at the results) to get all the details. Now I'll go get some sleep...
Ohh, before I leave, one more note: I have no way at all to test this code, so it's quite possible I've made mistakes in the patch I'm attaching, so do review it carefully before applying bits from it.
Signature · 1 line
Jesper Juhl <[email protected]>
We really don't have plan about Blackfin with a MMU now. So do you
beleve it is necessary to change the name to arch/blackfin?
Quoted 18 lines · Robert Schwebel
Quoted 2 lines · Luke Yang
This patch mainly includes the
arch/bfinnommuarchitecture files and some blackfin specific drivers.Having in mind the pain with arm and m68k with and without MMU, could you structure that thing in a way that it will be able to share the current nommu code with (hopefully coming) code for blackfins which might have a MMU? So it would be something like
arch/blackfin.Robert
--
Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
Pengutronix - Linux Solutions for Science and Industry
Handelsregister: Amtsgericht Hildesheim, HRA 2686
Hannoversche Str. 2, 31134 Hildesheim, Germany
Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9
Quoted 2 lines · Luke Yang
We really don't have plan about Blackfin with a MMU now. So do you beleve it is necessary to change the name to
arch/blackfin?
Hmm, as far as I know there are at least plans for variants with memory protection units (but without virtual addressing), but anyway - why put the mmu into the architecture name? If you use "blackfin" you are open to everything and no non dsp guru has to guess what bf might be ;)
Robert
Signature · 5 lines
Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
Pengutronix - Linux Solutions for Science and Industry
Handelsregister: Amtsgericht Hildesheim, HRA 2686
Hannoversche Str. 2, 31134 Hildesheim, Germany
Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9