How Many Kernel Branches For A Single Device, for example Tosa 6000L Options

[xjqian]

Please help me to understand the current kernel status for Tosa (or other Z breed):

pdaXrom: 2.4.18 and 2.6.xx

OZ: 2.4.18 (old) and 2.6.18

along the line of sharp compatible rom

Cacko: 2.4.18
Tetsu special kernel: 2.4.18
guyhelm: 2.4.18 (a.k.a. common kernel)

for the various 2.4.18 kernels, are most of patches same or equivalent? Are they all stemming from a common well patched version? Is it worth to dig into each one to find the best patch available or am I wasting my time?

[speculatrix]

As I understand it, it's more the case of which kernel for which machine.

The 760/860 with Sharp ROM (and thus Cacko) came with 2.4.18.

The 6000 had a variant of the 2.4.18 kernel as that model used the Toshiba "companion" chip for video and SD/MMC. That Tosh chip is mostly undocumented (presumably because SD was a secret standard at the time) and remains the stumbling block to getting accelerated video.


The x000 series came from Sharp with 2.4.20.

Older builds of OZ (and a relatively recent one for 6000) used 2.4 but all the recent releases use 2.6

I think stable versions of pdaXrom use 2.4.18 or 2.4.20 (much as Cacko does), but some versions in testing for some hardware use 2.6, but I don't know to be sure. Note that pdaXrom and OZ builds are incompatible because one uses hard float, the other soft float; I can't remember which is which.

Angstrom uses 2.6 (being, I suppose in some ways a fork of OZ) but uses EABI, which makes it completely software incompatible with anything else, but is a much better fit with the Arm architecture and is faster - by how much I don't know!

Please do correct me. Perhaps this should be put in the wiki once all amendments have been done.

-- edit --
found a wiki page had already been started: http://www.oesf.org/index.php?title=Kernels

[xjqian]

thanks for the reply. Maybe i didn't make myself clear. I understand the legacy 2.4.18 and 2.4.20 kernels are device specific. I also understand the 2.6.x kernels pdaXrom and OZ team are working on probably differ a lot. What I don't understand is how much the different 2.4.x kernels patches overlap. I'm very keen to the a common 2.4.18 kernel for Tosa (as suggested by guylhem), but I don't know how does that relate to the kernel used by pdaXrom beta1.

In summary, I'm asking what's the relationship, if any, between the different 2.4.18 kernels (more precisely, kernel patches) for Tosa or different 2.4.20 kernels (more precisely, kernel patches) for 3x00.

[Meanie]

thanks for the reply. Maybe i didn't make myself clear. I understand the legacy 2.4.18 and 2.4.20 kernels are device specific. I also understand the 2.6.x kernels pdaXrom and OZ team are working on probably differ a lot. What I don't understand is  how much the different 2.4.x kernels patches overlap. I'm very keen to the a common 2.4.18 kernel for Tosa (as suggested by guylhem), but I don't know how does that relate to the kernel used by pdaXrom beta1.

In summary, I'm asking what's the relationship, if any, between the different 2.4.18 kernels (more precisely, kernel patches) for Tosa or different 2.4.20 kernels  (more precisely, kernel patches) for 3x00.

anybody that compiled kernels just adds all the relevant patches that they can find at tne time of compile so depending on when the kernels were compiled the number of available patches might have differed and also "relevant" is subjective and depends on the person who compiles the kernel on whether he/she thinks that the patch is needed or not. In addition, there are many options when compiling a kernel and depending on who compiled the kernel, those options may also differ, ie CDROM/DVD support? some may say, the Z dun have no CD drive, thus no, but others may include it since they could foresee a USB host with CD drive....

[adf]

thanks for the reply. Maybe i didn't make myself clear. I understand the legacy 2.4.18 and 2.4.20 kernels are device specific. I also understand the 2.6.x kernels pdaXrom and OZ team are working on probably differ a lot. What I don't understand is  how much the different 2.4.x kernels patches overlap. I'm very keen to the a common 2.4.18 kernel for Tosa (as suggested by guylhem), but I don't know how does that relate to the kernel used by pdaXrom beta1.

In summary, I'm asking what's the relationship, if any, between the different 2.4.18 kernels (more precisely, kernel patches) for Tosa or different 2.4.20 kernels  (more precisely, kernel patches) for 3x00.

I think you are going to have a problem with the toshiba drivers on 2.4 for Tosa-- they are a big reason to stay with the hardfloat sharprom stuff. In thinking this over, I wonder if it might not be a better idea for tosa to get some 2.6 kernel sources and start working on a beta4 based stable? On the x000 clamshells much can be made of 2.4.20 with some pxa 270 optimizations. None of that is true for tosa. In fact, the easiest route for big sd support is to go 2.6, I think. Running OZ on my Tosa gave me better speed, and no lack of hardware support that I noticed.(you might want to ask around about that, though) Maybe going pdaX via 2.6 on Tosa is really the path of least resistance?

[speculatrix]

I think.  Running OZ on my Tosa gave me better speed, and no lack of hardware support that I noticed.(you might want to ask around about that, though)  Maybe going pdaX via 2.6 on Tosa is really the path of least resistance?

when I last tried OZ with k2.6 on tosa, I got the orange "residue" on the screen when images moved; I think it was a bug in the X11 driver, and was due to be solved - Sharp did some sort of hack in their driver to make it work?

[xjqian]

I think you are going to have a problem with the toshiba drivers on 2.4 for Tosa--  they are a big reason to stay with the hardfloat sharprom stuff.  In thinking this over, I wonder if it might not be a better idea for tosa to get some 2.6 kernel sources and start working on a beta4 based stable?  On the x000 clamshells much can be made of 2.4.20 with some pxa 270 optimizations.  None of that is true for tosa. In fact, the easiest route for big sd support is to go 2.6, I think.  Running OZ on my Tosa gave me better speed, and no lack of hardware support that I noticed.(you might want to ask around about that, though)  Maybe going pdaX via 2.6 on Tosa is really the path of least resistance?

hrw is working on the 2.6 kernel OZ release. So the 2.6 kernel Tosa is in good hands. I think an updated stable 2.4 kernel Tosa is still valuable. Besides, I have to learn to walk before run. I'm in the stage of collecting different 2.4.x kernel sources now.

[xjqian]

anybody that compiled kernels just adds all the relevant patches that they can find at tne time of compile so depending on when the kernels were compiled the number of available patches might have differed and also "relevant" is subjective and depends on the person who compiles the kernel on whether he/she thinks that the patch is needed or not. In addition, there are many options when compiling a kernel and depending on who compiled the kernel, those options may also differ, ie CDROM/DVD support? some may say, the Z dun have no CD drive, thus no, but others may include it since they could foresee a USB host with CD drive....

Thanks for the clear explanation. I read your pdaXii13 page. One thing puzzles me.

You said you tried Tetsu kernel, which works equally well as the kernel in pdaX beta1/3.

Presumably, the Tetsu kernel is hard float. The following maybe related as well.

You've also mentioned kernel/modules has to be built using gcc 2.95.

Does that mean the the kernel/modules are always hard float? The soft float only concerns applications built by gcc 3.4.x with sashz's special patch?

[Meanie]

Does that mean the the kernel/modules are always hard float? The soft float only concerns applications built by gcc 3.4.x with sashz's special patch?

correct. at least for the 2.4.x kernel stuff

[Hrw]

Tosa with 2.6.17 kernel works very nice. To use newer versions someone need to update Toshiba USB Host driver.

Orange artifacts are common in most of non-Sharp kernels. Someone in OE bugtracker wrote that this is a matter of proper aligning of few structures on each write.

Under 2.6 there is no way to disable USB WiFi port.

[adf]

the artifacting seems to have become less of an issue, though. There is still some the last OZ, but not nearly the problem it had been.

[xjqian]

correct. at least for the 2.4.x kernel stuff

I guess the rationale behind that is to keep compatible with the sharp proprietary SD/MMC driver (please advice if there's anything else)? Since the SD driver seems no longer a problem in 2.6 kernel, does that mean the whole kernel/module could be compiled by soft float enabled gcc? Is this the direction pdaXrom team are heading to?

[xjqian]

Tosa with 2.6.17 kernel works very nice. To use newer versions someone need to update Toshiba USB Host driver.

Orange artifacts are common in most of non-Sharp kernels. Someone in OE bugtracker wrote that this is a matter of proper aligning of few structures on each write.

Under 2.6 there is no way to disable USB WiFi port.

hrw: could you comment on if anything has been done to merge guylhem's "common kernel" into OZ? could you also give me a pointer to where I can download the latest (though old) OZ 2.4 kernel (which release)?

[adf]

Tosa with 2.6.17 kernel works very nice. To use newer versions someone need to update Toshiba USB Host driver.

Orange artifacts are common in most of non-Sharp kernels. Someone in OE bugtracker wrote that this is a matter of proper aligning of few structures on each write.

Under 2.6 there is no way to disable USB WiFi port.

hrw: could you comment on if anything has been done to merge guylhem's "common kernel" into OZ? could you also give me a pointer to where I can download the latest (though old) OZ 2.4 kernel (which release)?

I had meant the latest 2.6 kernel, If what I said is sending you off to the OZ 2.4. I think, If you go that route, you might get both the orange lines and still be stuck with the 2.4.18 sd driver for only <1gig sd.

My personal experience with Oz has been that the 2.6.17 was far superior to the 2.4 based version

[xjqian]

I had meant the latest 2.6 kernel, If what I said is sending you off to the OZ 2.4.  I think, If you go that route, you might get both the orange lines and still be stuck with the 2.4.18 sd driver for only <1gig sd.

My personal experience with Oz has been that the 2.6.17 was far superior to the 2.4 based version

No, you are not. I know you mean the 2.6 kernel. However, I will wait for OZ 3.5.5 to test again. The currently 3.5.4 rc2 release is not stable for me.

SD> 1g is desirable but not critical for Tosa compared with other Clamshells without integrated wifi. Instead, stability IS a big issue for me (one of the major reasons I want to get a second Tosa). We need people working on the cutting edge. But I'm trying to work on an updated stable 2.4 kernel release for Tosa.

I'm asking OZ team the 2.4 kernel, not because I'm optimistic about it, but because I'd like to know what's the status of 2.4 kernel at different branches. Since I'm not involved in development previously, I'd like to bring myself to the same page as other knowledgeable people in the forum.