[Letux-kernel] TILER patch problem on omap3 mmap(/dev/fb0) -- [PATCH] fbpng: make it work with 32 bpp fb

[H. Nikolaus Schaller]

Hi Andreas,

> Am 23.01.2018 um 23:09 schrieb Andreas Kemnade <andreas at kemnade.info>:
> 
> On Tue, 23 Jan 2018 10:38:13 +0100
> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
> [...]
>> 
>> I found a simpler "proof of bug" using the latest kernel from our downloads:
>> 
> I was wondering why you did not do that in the first place. I was just
> doing research about how to boot test setups, so I did it the other way.

Well, there is no need to wonder that I have different priorities
and available time for each topic is limited...

> 
> [...]
>> root at letux:~# ./fbpng /usr/share/pixmaps/debian-logo.png
>> /usr/share/pixmaps/debian-logo.png: 48x48 @ 24bp[  146.310180] Unhandled fault: imprecise external abort (0x1c06) at 0xb6c8d020
>> [  146.321441] pgd = e15e8173
>> [  146.324279] [b6c8d020] *pgd=958d2831, *pte=9e900353, *ppte=9e900873
>> p
>> [  146.331054] Unhandled fault: imprecise external abort (0x1c06) at 0xdc91bfe8
>> [  146.338592] pgd = 4405b38c
>> [  146.341430] [dc91bfe8] *pgd=9c80041e(bad)
>> Bus error
>> root at letux:~# 
>> 
>> On omap5 the same procedure succeeds.
>> 
>> Now I can help to find out what the reason could be...
>> 
> well, suspicious should be everything not guarded by if
> (... OMAP_BO_TILED ...).
> But for the other things, I do not understand so well whether that
> can have influences.

Me too - but I think it is not very difficult to pinpoint,
i.e. remove one patch after the other. Or bisect. The total
set of patches is not too big.

And experiment with some if().

BR,
Nikolaus