Hi Simon,
Simon Glass wrote:
On Wed, 31 Jul 2024 at 04:14, Chris Webb wrote:
Alas I don't have any boards to test on for either of these platforms.
If you have the inclination it is still worth sending a patch. The
maintainer can check it. These sorts of counter-examples can be copie
Hi Simon,
Simon Glass wrote:
Presumably it needs to apply to every mtk soc that uses
mtk_pinctrl_common_probe() as they'll all be affected by this problem.
Yes I suppose so.
As well as the mediatek case (patch just sent), I thought I should look
through the other pinctrl drivers for oth
() to mtk_pinctrl_common_bind() and call
this as the .bind method of each of the mediatek pinctrl drivers.
Signed-off-by: Chris Webb
---
drivers/pinctrl/mediatek/pinctrl-mt7622.c | 1 +
drivers/pinctrl/mediatek/pinctrl-mt7623.c | 1 +
drivers/pinctrl/mediatek/pinctrl-mt7629.c | 1
Hi Simon,
Simon Glass wrote:
Well, yes, mt7981_pinctrl is wrong since it is not actually binding
the GPIO devices until it itself is probed. It should do it when it is
bound.
Oh I see! Yes, I can see the mtk_gpiochip_register(dev) in
mtk_pinctrl_common_probe() exactly as you say.
Better
Simon Glass wrote:
We cannot probe devices when they are bound since it breaks the
ordering of driver model.
From your trace it looks like everything is happening after
relocation. I can't quite see what is actually going wrong. But if you
look at dm_init_and_scan(), it does the probe at the e
Chris Webb wrote:
Now the release is out, I'd be really keen to pick this one up and get
it fixed upstream if possible.
Hi Tom, is there anything more I can do to help out here? I'd love
upstream 2024.10 to ship with gpio-hog that works again.
Best wishes,
Chris.
Chris Webb wrote:
Tom Rini wrote:
Adding Marek, as the author of commit 48b3ecbedf82 ("gpio: Get rid of
gpio_hog_probe_all()").
Thanks! I don't claim this is the correct way to fix this, just that it
works.
Specifically, the two things I found that got gpio-hog wor
Tom Rini wrote:
Adding Marek, as the author of commit 48b3ecbedf82 ("gpio: Get rid of
gpio_hog_probe_all()").
Thanks! I don't claim this is the correct way to fix this, just that it
works.
Specifically, the two things I found that got gpio-hog working were
(a) adding an explicit probe
:
[...]
CPU: MediaTek MT7981
Model: GL.iNet GL-X3000
DRAM: 512 MiB
[...]
Core: 34 devices, 14 uclasses, devicetree: separate
MMC:
mmc@1123: 0
[...]
Probe them directly in gpio_post_bind instead.
Signed-off-by: Chris Webb
---
drivers/gpio/gpio-uclass.c | 4 +++-
1 file
Chris Webb wrote:
This shows gpio_post_bind being called, then the four gpio-hogs found
and set to probe after bind, but the gpio_hog_probe function is never
actually called. Presumably this is the problem - if they're not probed
then they'll never take effect?
I enabled CONF
I'm building u-boot for a mt7981a board, very similar to the in-tree
mt7981-emmc-rfb. For reference, my device tree and output of make
savedefconfig are here:
https://gist.github.com/arachsys/98048980728c652bea641b42cfd41a40
Everything boots fine, but I have a handful of gpios which need t
Simon Glass wrote:
Chris Webb wrote:
My worry about using binman for cracking such a simple nut is the
dependencies it pulls in that wouldn't otherwise be needed: the
documentation suggests at least lzma-alone, liblz4, and pyelftools.
I should probably update this. Unless yo
Mark Kettenis wrote:
[Simon Glass wrote:]
Well, for producing an image that works. E.g. producing a full image
(TPL, SPL and U-Boot) for an SD card - it would be nice to have
u-boot-rockchip.sd.bin or something like that.
True. That'd be an image you write at certain offset I suppose s
Matwey V. Kornilov wrote:
Here, we introduce generic u-boot-tpl-with-spl.img target which is the
TPL image followed by the SPL binary.
Having built U-Boot TPL + SPL images over and over again for testing on
various rk3399 hardware, something like this would certainly be a
convenience. It
xieqin...@gmail.com wrote:
From: Nick Xie
Add devicetree support for Khadas Edge-Captain.
Khadas Captain is the carrier board for Khadas Edge.
Tested-by: Chris Webb
Here's the boot log of TPL -> SPL -> U-Boot proper -> Linux on a 4GB Edge
board in a Captain V12 carrier,
xieqin...@gmail.com wrote:
From: Nick Xie
Add devicetree support for Khadas Edge.
Khadas Edge is an expandable Rockchip RK3399 board with goldfinger.
Tested-by: Chris Webb
Here's the boot log of TPL -> SPL -> U-Boot proper -> Linux on a 4GB Edge
board booting from an
Nick Xie wrote:
That's great! I'll update the patches and send them soon.
I'll make sure I test your specific patch when you post it, but I can
already confirm that u-boot.git master happily boots a Khadas Edge board.
I just added the unmodified Edge device tree from mainline Linux into
Hi Nick. I think Kever has merged the LPDDR4 series, and it's already made
its way into the mainline u-boot master branch.
https://gitlab.denx.de/u-boot/u-boot/commit/852f6ddd76fad2d5adef3f7e3a75d0065c68db3b
and its ancestors are the v3 series Jagan posted to the list.
There have also been
The generic code in common/spl/spl.c allows TPL/SPL banners to be
silenced by unsetting CONFIG_TPL_BANNER_PRINT or CONFIG_SPL_BANNER_PRINT
respectively. However, arch/arm/mach-rockchip/tpl.c prints this banner
unconditionally.
Fix the rockchip-specific tpl.c so that the TPL banner depends on
CONFI
If CONFIG_DEBUG_UART is set but CONFIG_TPL_SERIAL_SUPPORT is not, the
serial output should be available in SPL and full U-Boot, but not built
in TPL. However, the rockchip tpl.c instead fails to compile with
undefined references to the debug UART.
Instead, initialise the debug UART and print the T
These are two trivial patches which fix up the #ifdef conditionals in
mach-rockchip/tpl.c to do the right thing in two configuration scenarios:
1. Debug UART enabled (for SPL and main U-Boot) but serial support disabled
for the TPL stage.
2. TPL banner disabled by unsetting CONFIG_TPL_BANNER_PRIN
make_fit_atf.py no longer requires pyelftools, and nothing else in the
rockchip build requires it either, so remove references to installing it
from the documentation.
Signed-off-by: Chris Webb
---
board/rockchip/evb_rk3399/README | 6 --
doc/README.rockchip | 4
2 files
the destructuring code is reasonably
self-documenting.
Signed-off-by: Chris Webb
---
arch/arm/mach-rockchip/make_fit_atf.py | 75 +++---
1 file changed, 32 insertions(+), 43 deletions(-)
diff --git a/arch/arm/mach-rockchip/make_fit_atf.py
b/arch/arm/mach-rockchip
Building for rockchip, make_fit_atf.py depends on pyelftools, a non-bundled
python module that is not used elsewhere in building u-boot or the kernel.
We only use pyelftools to pull out PT_LOAD segments. ELF is very simple, so
doing this manually is easy and spares users the extra dependency. In f
Hi Andy and Kever
Andy Yan writes:
> One small coding style issue:
>
> offset = e_phoff + e_phentsize * index
Thanks, will fix this.
Kever Yang writes:
> Please fix the coding style comment by Andy, for other codes:
>
> Reviewed-by: Kever Yang
Newbie contributor so can I check: is it rig
make_fit_atf.py no longer requires pyelftools, and nothing else in the
rockchip build requires it either, so remove references to installing it
from the documentation.
Signed-off-by: Chris Webb
---
board/rockchip/evb_rk3399/README | 6 --
doc/README.rockchip | 4
2 files
the destructuring code is reasonably
self-documenting.
Signed-off-by: Chris Webb
---
arch/arm/mach-rockchip/make_fit_atf.py | 75 +++---
1 file changed, 32 insertions(+), 43 deletions(-)
diff --git a/arch/arm/mach-rockchip/make_fit_atf.py
b/arch/arm/mach-rockchip
[Sorry for a respin so soon, but this fixes the overzealous documentation
update and also ensures the script's behaviour is identical to the existing
implementation in the case of a completely empty bl31 ELF file.]
Building for Rockchip, make_fit_atf.py depends on pyelftools, a non-bundled
python
Oops, sorry; I'm an idiot. The synopsys board does indeed still use
pyelftools, and therefore while these are correct:
> board/rockchip/evb_rk3399/README | 6 --
> doc/README.rockchip | 4
these removals are incorrect:
> .travis.yml | 1 -
> board/syno
make_fit_atf.py no longer requires pyelftools, and nothing else in the
tree requires it either, so remove references to installing it from the
documentation and travis configuration.
Signed-off-by: Chris Webb
---
.travis.yml | 1 -
board/rockchip/evb_rk3399/README | 6
the destructuring code is reasonably
self-documenting.
Signed-off-by: Chris Webb
---
arch/arm/mach-rockchip/make_fit_atf.py | 74 +++---
1 file changed, 31 insertions(+), 43 deletions(-)
diff --git a/arch/arm/mach-rockchip/make_fit_atf.py
b/arch/arm/mach-rockchip
Building for rockchip, make_fit_atf.py depends on pyelftools, a non-bundled
python module that is not used elsewhere in building u-boot or the kernel.
We only use pyelftools to pull out PT_LOAD segments. ELF is very simple, so
doing this manually is easy and spares users the extra dependency. In f
Jagan Teki writes:
> This is the v2 set for supporting LPDDR4 with associated features
Sorry for a late follow-up, but I've just tested this series (applied to
today's master branch) on Khadas Edge.
With these patches, a device tree copied from Linux -next, and a defconfig
essentially the same
33 matches
Mail list logo
