mtd: rawnand: brcmnand: exec_op patch broke < v5.0 controllers

mtd: rawnand: brcmnand: exec_op patch broke < v5.0 controllers

[Álvaro Fernández Rojas]

Hi David,

I've just realized that your exec_op implementation patch broke
support of v2.1, v2.2 and v4.0 controllers.
https://github.com/torvalds/linux/commit/3c8260ce76634291aed877032a41e373884d69e4
https://patchwork.ozlabs.org/project/linux-mtd/patch/20231125012438.15191-4-dregan@broadcom.com/
I know it's an old patch from 6.8 but I just discovered this while
trying to add 6.12 support for bmips on OpenWrt (from 6.6):
https://github.com/openwrt/openwrt/pull/18746

The issue is that low level ops always return 0 on v4.0 controllers
and v2.1/v2.2 controllers don't even have the low level registers
defined:
https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L357
https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L371

I guess that restoring the legacy functions isn't going to happen, but
is it possible to restore < v5.0 compatibility on the new exec_op
implementation?

Best regards,
Álvaro.

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

mtd: rawnand: brcmnand: exec_op patch broke < v5.0 controllers

[Álvaro Fernández Rojas]

Hi David,

I've just realized that your exec_op implementation patch broke
support of v2.1, v2.2 and v4.0 controllers.
https://github.com/torvalds/linux/commit/3c8260ce76634291aed877032a41e373884d69e4
https://patchwork.ozlabs.org/project/linux-mtd/patch/20231125012438.15191-4-dregan@broadcom.com/
I know it's an old patch from 6.8 but I just discovered this while
trying to add 6.12 support for bmips on OpenWrt (from 6.6):
https://github.com/openwrt/openwrt/pull/18746

The issue is that low level ops always return 0 on v4.0 controllers
and v2.1/v2.2 controllers don't even have the low level registers
defined:
https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L357
https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L371

I guess that restoring the legacy functions isn't going to happen, but
is it possible to restore < v5.0 compatibility on the new exec_op
implementation?

Best regards,
Álvaro.

Re: mtd: rawnand: brcmnand: exec_op patch broke < v5.0 controllers

[Florian Fainelli]

On 5/11/2025 11:42 AM, Álvaro Fernández Rojas wrote:
> Hi David,
> 
> I've just realized that your exec_op implementation patch broke
> support of v2.1, v2.2 and v4.0 controllers.
> https://github.com/torvalds/linux/commit/3c8260ce76634291aed877032a41e373884d69e4
> https://patchwork.ozlabs.org/project/linux-mtd/patch/20231125012438.15191-4-dregan@broadcom.com/
> I know it's an old patch from 6.8 but I just discovered this while
> trying to add 6.12 support for bmips on OpenWrt (from 6.6):
> https://github.com/openwrt/openwrt/pull/18746
> 
> The issue is that low level ops always return 0 on v4.0 controllers
> and v2.1/v2.2 controllers don't even have the low level registers
> defined:
> https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L357
> https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L371
> 
> I guess that restoring the legacy functions isn't going to happen, but
> is it possible to restore < v5.0 compatibility on the new exec_op
> implementation?

Unfortunately, I don't think we have any other choice unless we want to 
have a functional NAND driver for these older chips. Would you be 
willing to do that? Thanks
-- 
Florian


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: mtd: rawnand: brcmnand: exec_op patch broke < v5.0 controllers

[Florian Fainelli]

On 5/11/2025 11:42 AM, Álvaro Fernández Rojas wrote:
> Hi David,
> 
> I've just realized that your exec_op implementation patch broke
> support of v2.1, v2.2 and v4.0 controllers.
> https://github.com/torvalds/linux/commit/3c8260ce76634291aed877032a41e373884d69e4
> https://patchwork.ozlabs.org/project/linux-mtd/patch/20231125012438.15191-4-dregan@broadcom.com/
> I know it's an old patch from 6.8 but I just discovered this while
> trying to add 6.12 support for bmips on OpenWrt (from 6.6):
> https://github.com/openwrt/openwrt/pull/18746
> 
> The issue is that low level ops always return 0 on v4.0 controllers
> and v2.1/v2.2 controllers don't even have the low level registers
> defined:
> https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L357
> https://github.com/torvalds/linux/blob/3ce9925823c7d6bb0e6eb951bf2db0e9e182582d/drivers/mtd/nand/raw/brcmnand/brcmnand.c#L371
> 
> I guess that restoring the legacy functions isn't going to happen, but
> is it possible to restore < v5.0 compatibility on the new exec_op
> implementation?

Unfortunately, I don't think we have any other choice unless we want to 
have a functional NAND driver for these older chips. Would you be 
willing to do that? Thanks
-- 
Florian