Skip to contentSkip to conversation

fdtgrep exclude nodes

U-Boot Michal Simek 3 participants 6 messages Dec 1, 2015

lore
Share & tools
lore
#0 lore raw

Hi Simon and Stefan,

I was checking spl/u-boot-spl.dtb and I see that you are removing interrupt-parent property but I think that also make sense to remove interrupt property as well because it is unused and parent is also removed. Just extend OF_SPL_REMOVE_PROPS by "interrupts" to save some memory.

Another thing I was checking was that we are missing rules for removing nodes with status = "disabled" property. I think that will be good to add it. The reason is that I would move all that u-boot,dm-pre-reloc to dtsi at least for now.
Or do you also probe drivers for nodes which have status disabled?

Thanks,
Michal

Signature · 14 lines
Michal Simek, Ing. (M.Eng), OpenPGP -> KeyID: FE3D1F91
w: www.monstr.eu p: +42-0-721842854
Maintainer of Linux kernel - Microblaze cpu - http://www.monstr.eu/fdt/
Maintainer of Linux kernel - Xilinx Zynq ARM architecture
Microblaze U-BOOT custodian and responsible for u-boot arm zynq platform


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: OpenPGP digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20151201/d2def96b/attachment.sig>
#1 lore raw

Hi Michal,

Quoted 4 lines · Michal Simek

I was checking spl/u-boot-spl.dtb and I see that you are removing interrupt-parent property but I think that also make sense to remove interrupt property as well because it is unused and parent is also removed. Just extend OF_SPL_REMOVE_PROPS by "interrupts" to save some memory.

I'm fine with this.

Quoted 4 lines

Another thing I was checking was that we are missing rules for removing nodes with status = "disabled" property. I think that will be good to add it. The reason is that I would move all that u-boot,dm-pre-reloc to dtsi at least for now.

Hmmm. I prefer to add this "u-boot,dm-pre-reloc" to the board dts file instead of using the dtsi files. At least for the real devices, like UART etc. As they vary between the different boards.

Perhaps its possible to enable those busses (simple-bus) needed to traverse to the requested device automatically via fdtgrep? That would make this much easier and less error-prone. And would result in less "u-boot,dm-pre-reloc" occurances in the dts / dtsi files.

Thanks,
Stefan

#2 lore raw
Quoted 11 lines · Stefan Roese

Hi Michal,

Quoted 5 lines · Michal Simek

I was checking spl/u-boot-spl.dtb and I see that you are removing interrupt-parent property but I think that also make sense to remove interrupt property as well because it is unused and parent is also removed.
Just extend OF_SPL_REMOVE_PROPS by "interrupts" to save some memory.

I'm fine with this.

Quoted 8 lines
Quoted 4 lines

Another thing I was checking was that we are missing rules for removing nodes with status = "disabled" property. I think that will be good to add it. The reason is that I would move all that u-boot,dm-pre-reloc to dtsi at least for now.

Hmmm. I prefer to add this "u-boot,dm-pre-reloc" to the board dts file instead of using the dtsi files. At least for the real devices, like UART etc. As they vary between the different boards.

How do you handle if you have 2 uarts which one is primary one? That only one has dm-pre-reloc? Or are you using aliases?

Quoted 5 lines

Perhaps its possible to enable those busses (simple-bus) needed to traverse to the requested device automatically via fdtgrep? That would make this much easier and less error-prone. And would result in less "u-boot,dm-pre-reloc" occurances in the dts / dtsi files.

I still think that instead of putting this property to every node having one property with list of devices in chosen node is better way. But I haven't had a time to play with it.

Thanks,
Michal

Signature · 14 lines
Michal Simek, Ing. (M.Eng), OpenPGP -> KeyID: FE3D1F91
w: www.monstr.eu p: +42-0-721842854
Maintainer of Linux kernel - Microblaze cpu - http://www.monstr.eu/fdt/
Maintainer of Linux kernel - Xilinx Zynq ARM architecture
Microblaze U-BOOT custodian and responsible for u-boot arm zynq platform


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: OpenPGP digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20151201/94d02572/attachment.sig>
#3 lore raw
Quoted 17 lines · Michal Simek
Quoted 16 lines · Stefan Roese

Hi Michal,

Quoted 5 lines · Michal Simek

I was checking spl/u-boot-spl.dtb and I see that you are removing interrupt-parent property but I think that also make sense to remove interrupt property as well because it is unused and parent is also removed.
Just extend OF_SPL_REMOVE_PROPS by "interrupts" to save some memory.

I'm fine with this.

Quoted 4 lines

Another thing I was checking was that we are missing rules for removing nodes with status = "disabled" property. I think that will be good to add it. The reason is that I would move all that u-boot,dm-pre-reloc to dtsi at least for now.

Quoted 6 lines
Quoted 3 lines

Hmmm. I prefer to add this "u-boot,dm-pre-reloc" to the board dts file instead of using the dtsi files. At least for the real devices, like UART etc. As they vary between the different boards.

How do you handle if you have 2 uarts which one is primary one? That only one has dm-pre-reloc? Or are you using aliases?

I only mark the devices absolutely necessary in SPL for booting with dm-pre-reloc. And that only one UART, one SPI NOR etc. And yes, I'm also using aliases.

Quoted 9 lines
Quoted 6 lines

Perhaps its possible to enable those busses (simple-bus) needed to traverse to the requested device automatically via fdtgrep? That would make this much easier and less error-prone. And would result in less "u-boot,dm-pre-reloc" occurances in the dts / dtsi files.

I still think that instead of putting this property to every node having one property with list of devices in chosen node is better way.

Yes, that would also be a good solution. Sprinkling these pre-reloc properties all over the files and nodes / busses is definitely not perfect.

Thanks,
Stefan

#4 lore raw

Hi Stefan,

Quoted 44 lines · Stefan Roese
Quoted 22 lines · Michal Simek
Quoted 20 lines · Stefan Roese

Hi Michal,

Quoted 6 lines · Michal Simek

I was checking spl/u-boot-spl.dtb and I see that you are removing interrupt-parent property but I think that also make sense to remove interrupt property as well because it is unused and parent is also removed.
Just extend OF_SPL_REMOVE_PROPS by "interrupts" to save some memory.

I'm fine with this.

Quoted 4 lines

Another thing I was checking was that we are missing rules for removing nodes with status = "disabled" property. I think that will be good to add it. The reason is that I would move all that u-boot,dm-pre-reloc to dtsi at least for now.

Quoted 7 lines
Quoted 3 lines

Hmmm. I prefer to add this "u-boot,dm-pre-reloc" to the board dts file instead of using the dtsi files. At least for the real devices, like UART etc. As they vary between the different boards.

How do you handle if you have 2 uarts which one is primary one? That only one has dm-pre-reloc? Or are you using aliases?

I only mark the devices absolutely necessary in SPL for booting with dm-pre-reloc. And that only one UART, one SPI NOR etc. And yes, I'm also using aliases.

Quoted 8 lines
Quoted 6 lines

Perhaps its possible to enable those busses (simple-bus) needed to traverse to the requested device automatically via fdtgrep? That would make this much easier and less error-prone. And would result in less "u-boot,dm-pre-reloc" occurances in the dts / dtsi files.

Quoted 7 lines
Quoted 2 lines

I still think that instead of putting this property to every node having one property with list of devices in chosen node is better way.

Yes, that would also be a good solution. Sprinkling these pre-reloc properties all over the files and nodes / busses is definitely not perfect.

This has come up before.

I can see pros and cons, but it's not clear to me that one way is better than the other. I am concerned that we use the 'best' way to avoid increasing amounts of rework later if we change the approach.

If someone is willing to sketch out how it would be implemented with a single property (e.g. on a new thread) then that would help continue the discussion.

Regards,
Simon

#5 lore raw
In reply to Stefan Roese
Quoted 22 lines · Stefan Roese
Quoted 21 lines · Michal Simek
Quoted 19 lines · Stefan Roese

Hi Michal,

Quoted 5 lines · Michal Simek

I was checking spl/u-boot-spl.dtb and I see that you are removing interrupt-parent property but I think that also make sense to remove interrupt property as well because it is unused and parent is also removed.
Just extend OF_SPL_REMOVE_PROPS by "interrupts" to save some memory.

I'm fine with this.

Quoted 4 lines

Another thing I was checking was that we are missing rules for removing nodes with status = "disabled" property. I think that will be good to add it. The reason is that I would move all that u-boot,dm-pre-reloc to dtsi at least for now.

Hmmm. I prefer to add this "u-boot,dm-pre-reloc" to the board dts file instead of using the dtsi files. At least for the real devices, like UART etc. As they vary between the different boards.

Quoted 6 lines
Quoted 2 lines

How do you handle if you have 2 uarts which one is primary one? That only one has dm-pre-reloc? Or are you using aliases?

I only mark the devices absolutely necessary in SPL for booting with dm-pre-reloc. And that only one UART, one SPI NOR etc. And yes, I'm also using aliases.

TBH I tend to add dm-pre-reloc to DTSI for all IPs which we can use in SPL. It means by default all devices which can be used for booting will be there. In combination of removing all IPs which have status="disabled" I think I am getting pretty accurate picture of all devices which SPL needs.
There could be small penalty for dt size, SPL size and maybe somewhere else but I think it will pretty much solve all problems regarding missing dm-pre-reloc somewhere.
If someone wants to have own production board, they can remove these dm-pre-reloc properties to shorten time and having smaller footprint.

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

Perhaps its possible to enable those busses (simple-bus) needed to traverse to the requested device automatically via fdtgrep? That would make this much easier and less error-prone. And would result in less "u-boot,dm-pre-reloc" occurances in the dts / dtsi files.

Quoted 6 lines
Quoted 2 lines

I still think that instead of putting this property to every node having one property with list of devices in chosen node is better way.

Yes, that would also be a good solution. Sprinkling these pre-reloc properties all over the files and nodes / busses is definitely not perfect.

ok.

Thanks,
Michal

Signature · 14 lines
Michal Simek, Ing. (M.Eng), OpenPGP -> KeyID: FE3D1F91
w: www.monstr.eu p: +42-0-721842854
Maintainer of Linux kernel - Microblaze cpu - http://www.monstr.eu/fdt/
Maintainer of Linux kernel - Xilinx Zynq ARM architecture
Microblaze U-BOOT custodian and responsible for u-boot arm zynq platform


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: OpenPGP digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20151201/35afb7f7/attachment.sig>