Re: [PATCH v3 2/2] arm64: dts: Add BRCM IPROC NAND DT node for NS2

From: Ray Jui
Date: Wed Oct 28 2015 - 12:08:11 EST

On 10/28/2015 2:06 AM, Anup Patel wrote:

-----Original Message-----
From: Ray Jui [mailto:rjui@xxxxxxxxxxxx]
Sent: 28 October 2015 06:17
To: Brian Norris
Cc: Anup Patel; David Woodhouse; Linux MTD; Rob Herring; Pawel Moll; Mark
Rutland; Catalin Marinas; Will Deacon; Sudeep Holla; Ian Campbell; Kumar Gala;
Scott Branden; Florian Fainelli; Pramod Kumar; Vikram Prakash; Sandeep
Tripathy; Linux ARM Kernel; Device Tree; Linux Kernel; bcm-kernel-feedback-list
Subject: Re: [PATCH v3 2/2] arm64: dts: Add BRCM IPROC NAND DT node for

On 10/27/2015 5:39 PM, Brian Norris wrote:
On Tue, Oct 27, 2015 at 05:25:32PM -0700, Ray Jui wrote:
On 10/27/2015 5:19 PM, Brian Norris wrote:
On Fri, Oct 23, 2015 at 10:46:13AM +0530, Anup Patel wrote:
diff --git a/arch/arm64/boot/dts/broadcom/ns2.dtsi
index f603277..9610822 100644
--- a/arch/arm64/boot/dts/broadcom/ns2.dtsi
+++ b/arch/arm64/boot/dts/broadcom/ns2.dtsi
@@ -212,5 +212,19 @@
compatible = "brcm,iproc-rng200";
reg = <0x66220000 0x28>;
+ nand: nand@66460000 {
+ compatible = "brcm,nand-iproc", "brcm,brcmnand-

Technically, the binding says you should also have "brcm,brcmnand"
as a last resort. Otherwise (for the NAND parts):

I believe Anup was seeing issues when both "brcm,nand-iproc" and
"brcm,brcmnand" are present.

Note "brcm,nand-iproc" invokes 'iproc_nand_probe', which calls
'brcmnand_probe' in the end.

"brcm,brcmnand" invokes 'brcmstb_nand_probe', which also calls
'brcmstb_probe', but without all the prep configuration required for

Ah, I forgot about that problem. That seems like an OF infrastructure
issue that could be fixed. We could lump these drivers back together,
and make sure that "brcm,nand-iproc" gets the priority in the
of_device_id list.

Or we could just relax the DT binding.

But wait, wouldn't cygnus already have that problem? You're using the
binding I suggested in arch/arm/boot/dts/bcm-cygnus.dtsi.

Interestingly, we do not see this problem with Cygnus or NSP, but only on NS2
(arm64 based). There may be a difference between how OF devices are
instantiated between arm and arm64?

Alternately, it could be also about order in-which platform drivers are matched
for newly created OF device.

Oh, and I see we hacked this one in drivers/mtd/nand/brcmnand/Makefile:

# link order matters; don't link the more generic brcmstb_nand.o before the
# more specific iproc_nand.o, for instance

Yes, I see that too (after sending out my previous email, :)). Maybe
Anup can help to elaborate on the problem. I'm now getting a bit
confused on how the problem can surface on NS2.

I think for a newly created OF devices the Linux device driver framework will
match the platform drivers in the order in which they are registered by module
init functions. Now the order of module init function calls will be based how
the .initcall section is formed by linker and order in which objects are linked.

Yes, what you said is my understanding as well, but then here is the mystery. This is the link order in brcmnand/Makefile:

1 # link order matters; don't link the more generic brcmstb_nand.o before the
2 # more specific iproc_nand.o, for instance
3 obj-$(CONFIG_MTD_NAND_BRCMNAND) += iproc_nand.o
4 obj-$(CONFIG_MTD_NAND_BRCMNAND) += bcm63138_nand.o
5 obj-$(CONFIG_MTD_NAND_BRCMNAND) += brcmstb_nand.o
6 obj-$(CONFIG_MTD_NAND_BRCMNAND) += brcmnand.o

Based on the order above, probe from iproc_nand should always be called first if a matching compatible string is found. If so, then why having both compatible strings "brcm,brcmnand" and "brcm,nand-iproc" causes issues for NS2 (I remember it broke smoketest in the past when you submitted the change)? I'm not saying we should have "brcm,brcmnand" for iProc devices, but I don't get why it would cause any issue.

Does the order of the compatible string matter when they are assigned to the same 'compatible' property like this?

compatible = "brcm,nand-iproc", "brcm,brcmnand-v6.1", "brcm,brcmnand";

IMHO, if multiple platform drivers match given OF device then platform driver
with longest matching compatible string should only be probed. I don't know
how big change this would be for OF framework.

But in general, I think it's a good idea to relax the requirement in the
DT binding document to not require "brcm,brcmnand", in the case when
"brcm,nand-iproc" and "brcm,nand-bcm63138" are present.

Even I think, it will be good to relax the DT bindings requirement for
BRCM NAND driver.


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at