linux-next: build failure on power pc

Hi all,

I found the power pc build broken on today's
linux-next (647060f3b592).

My compiler is:

yury:linux$ powerpc-linux-gnu-gcc --version
powerpc-linux-gnu-gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

The config and error log are attached.

Thanks,
Yury

On 1/20/21 1:29 PM, Yury Norov wrote:

Hi all,

I found the power pc build broken on today's
linux-next (647060f3b592).

Darn, I was building linux-5.11-rc4.

I'll try linux-next after I send this.

---
From: Randy Dunlap <redacted>

Fulfill AKEBONO Kconfig requirements.

Fixes these Kconfig warnings (and more) and fixes the subsequent
build errors:

WARNING: unmet direct dependencies detected for NETDEVICES
  Depends on [n]: NET [=n]
  Selected by [y]:
  - AKEBONO [=y] && PPC_47x [=y]

WARNING: unmet direct dependencies detected for MMC_SDHCI
  Depends on [n]: MMC [=n] && HAS_DMA [=y]
  Selected by [y]:
  - AKEBONO [=y] && PPC_47x [=y]

Signed-off-by: Randy Dunlap <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: Yury Norov <yury.norov@gmail.com>
---
 arch/powerpc/platforms/44x/Kconfig |    2 ++
 1 file changed, 2 insertions(+)

--- lnx-511-rc4.orig/arch/powerpc/platforms/44x/Kconfig
+++ lnx-511-rc4/arch/powerpc/platforms/44x/Kconfig

@@ -206,6 +206,7 @@ config AKEBONO
 	select PPC4xx_HSTA_MSI
 	select I2C
 	select I2C_IBM_IIC
+	select NET
 	select NETDEVICES
 	select ETHERNET
 	select NET_VENDOR_IBM

@@ -213,6 +214,7 @@ config AKEBONO
 	select USB if USB_SUPPORT
 	select USB_OHCI_HCD_PLATFORM if USB_OHCI_HCD
 	select USB_EHCI_HCD_PLATFORM if USB_EHCI_HCD
+	select MMC
 	select MMC_SDHCI
 	select MMC_SDHCI_PLTFM
 	select ATA

On Wed, Jan 20, 2021 at 10:10 PM Randy Dunlap [off-list ref] wrote:

Looks working, thanks.

Tested-by: Yury Norov <yury.norov@gmail.com>

Randy Dunlap [off-list ref] writes:

I think the problem here is too much use of select, for things that
should instead be in the defconfig.

The patch below results in the same result for make
44x/akebono_defconfig. Does it fix the original issue?

We don't need to add ETHERNET or NET_VENDOR_IBM to the defconfig because
they're both default y.

cheers

diff --git a/arch/powerpc/configs/44x/akebono_defconfig b/arch/powerpc/configs/44x/akebono_defconfig
index 3894ba8f8ffc..6b08a85f4ce6 100644
--- a/arch/powerpc/configs/44x/akebono_defconfig
+++ b/arch/powerpc/configs/44x/akebono_defconfig

@@ -21,6 +21,7 @@ CONFIG_IRQ_ALL_CPUS=y
 # CONFIG_COMPACTION is not set
 # CONFIG_SUSPEND is not set
 CONFIG_NET=y
+CONFIG_NETDEVICES=y
 CONFIG_PACKET=y
 CONFIG_UNIX=y
 CONFIG_INET=y

@@ -98,6 +99,8 @@ CONFIG_USB_OHCI_HCD=y
 # CONFIG_USB_OHCI_HCD_PCI is not set
 CONFIG_USB_STORAGE=y
 CONFIG_MMC=y
+CONFIG_MMC_SDHCI=y
+CONFIG_MMC_SDHCI_PLTFM=y
 CONFIG_RTC_CLASS=y
 CONFIG_RTC_DRV_M41T80=y
 CONFIG_EXT2_FS=y

diff --git a/arch/powerpc/platforms/44x/Kconfig b/arch/powerpc/platforms/44x/Kconfig
index 78ac6d67a935..509b329c112f 100644
--- a/arch/powerpc/platforms/44x/Kconfig
+++ b/arch/powerpc/platforms/44x/Kconfig

@@ -206,15 +206,10 @@ config AKEBONO
        select PPC4xx_HSTA_MSI
        select I2C
        select I2C_IBM_IIC
-       select NETDEVICES
-       select ETHERNET
-       select NET_VENDOR_IBM
        select IBM_EMAC_EMAC4 if IBM_EMAC
        select USB if USB_SUPPORT
        select USB_OHCI_HCD_PLATFORM if USB_OHCI_HCD
        select USB_EHCI_HCD_PLATFORM if USB_EHCI_HCD
-       select MMC_SDHCI
-       select MMC_SDHCI_PLTFM
        select ATA
        select SATA_AHCI_PLATFORM
        help

On January 21, 2021 5:14:23 PM PST, Michael Ellerman [off-list ref] wrote:

Sure. I thought that lots of what was already there
should be in the defconfig. I  was just going with the flow. 

Thanks for fixing it. 

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

On 1/21/21 5:14 PM, Michael Ellerman wrote:

Randy Dunlap [off-list ref] writes:

quoted

On 1/20/21 1:29 PM, Yury Norov wrote:

quoted

Hi all,

I found the power pc build broken on today's
linux-next (647060f3b592).

Darn, I was building linux-5.11-rc4.

I'll try linux-next after I send this.

---
From: Randy Dunlap <redacted>

Fulfill AKEBONO Kconfig requirements.

Fixes these Kconfig warnings (and more) and fixes the subsequent
build errors:

WARNING: unmet direct dependencies detected for NETDEVICES
   Depends on [n]: NET [=n]
   Selected by [y]:
   - AKEBONO [=y] && PPC_47x [=y]

WARNING: unmet direct dependencies detected for MMC_SDHCI
   Depends on [n]: MMC [=n] && HAS_DMA [=y]
   Selected by [y]:
   - AKEBONO [=y] && PPC_47x [=y]

Signed-off-by: Randy Dunlap <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: Yury Norov <yury.norov@gmail.com>
---
  arch/powerpc/platforms/44x/Kconfig |    2 ++
  1 file changed, 2 insertions(+)

--- lnx-511-rc4.orig/arch/powerpc/platforms/44x/Kconfig
+++ lnx-511-rc4/arch/powerpc/platforms/44x/Kconfig

@@ -206,6 +206,7 @@ config AKEBONO
  	select PPC4xx_HSTA_MSI
  	select I2C
  	select I2C_IBM_IIC
+	select NET
  	select NETDEVICES
  	select ETHERNET
  	select NET_VENDOR_IBM

I think the problem here is too much use of select, for things that
should instead be in the defconfig.

The patch below results in the same result for make
44x/akebono_defconfig. Does it fix the original issue?

Hi Michael,
Sorry for the delay.

Changing the akebono_defconfig doesn't cause the missing symbols
to be set -- the defconfig is not being used here.

I guess that if you have users who set CONFIG_AKEBONO and expect
it to build cleanly, you will need something like my patch or the
patch that Florian just posted.

Changing the akebono_defconfig also would not help 'make randconfig'
builds to build cleanly if they had happened to enable AKEBONO.

We don't need to add ETHERNET or NET_VENDOR_IBM to the defconfig because
they're both default y.

cheers

Randy Dunlap [off-list ref] writes:

On 1/21/21 5:14 PM, Michael Ellerman wrote:

quoted

Randy Dunlap [off-list ref] writes:

quoted

On 1/20/21 1:29 PM, Yury Norov wrote:

quoted

Hi all,

I found the power pc build broken on today's
linux-next (647060f3b592).

Darn, I was building linux-5.11-rc4.

I'll try linux-next after I send this.

---
From: Randy Dunlap <redacted>

Fulfill AKEBONO Kconfig requirements.

Fixes these Kconfig warnings (and more) and fixes the subsequent
build errors:

WARNING: unmet direct dependencies detected for NETDEVICES
   Depends on [n]: NET [=n]
   Selected by [y]:
   - AKEBONO [=y] && PPC_47x [=y]

WARNING: unmet direct dependencies detected for MMC_SDHCI
   Depends on [n]: MMC [=n] && HAS_DMA [=y]
   Selected by [y]:
   - AKEBONO [=y] && PPC_47x [=y]

Signed-off-by: Randy Dunlap <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: Yury Norov <yury.norov@gmail.com>
---
  arch/powerpc/platforms/44x/Kconfig |    2 ++
  1 file changed, 2 insertions(+)

--- lnx-511-rc4.orig/arch/powerpc/platforms/44x/Kconfig
+++ lnx-511-rc4/arch/powerpc/platforms/44x/Kconfig

@@ -206,6 +206,7 @@ config AKEBONO
  	select PPC4xx_HSTA_MSI
  	select I2C
  	select I2C_IBM_IIC
+	select NET
  	select NETDEVICES
  	select ETHERNET
  	select NET_VENDOR_IBM

I think the problem here is too much use of select, for things that
should instead be in the defconfig.

The patch below results in the same result for make
44x/akebono_defconfig. Does it fix the original issue?

Hi Michael,
Sorry for the delay.

Changing the akebono_defconfig doesn't cause the missing symbols
to be set -- the defconfig is not being used here.

Yep, but that's OK. None of those selected symbols are hard dependencies
of AKEBONO, they're just things you probably want in your kernel to
actually boot on an akebono board.

I guess that if you have users who set CONFIG_AKEBONO and expect
it to build cleanly, you will need something like my patch or the
patch that Florian just posted.

It will build cleanly, it just won't necessarily boot on a real board.
Users who enable AKEBONO manually need to know what they're doing, or
they should just use the defconfig.

Changing the akebono_defconfig also would not help 'make randconfig'
builds to build cleanly if they had happened to enable AKEBONO.

Changing the defconfig doesn't help randconfig, but dropping the selects
does.

Anyway I'll send a proper version of my patch, which I'm pretty
confident will fix all the issues.

cheers