Skip to contentSkip to conversation

ACPI / LPSS: Make PCI dependency explicit

PATCH Superseded Sinan Kaya 11 participants 37 messages Jan 2, 2019 – Jan 11, 2019 1 message not in the archive

Subsystems (12)
DRM drivers x86 Sound DRM misc +8 more subsystems

v5 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Sinan Kaya

author
#1 lore raw

PATCH v5 1/11 ACPI / LPSS: Make PCI dependency explicit

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")', it is possible to build ACPI without any PCI support. This code depends on PCI. Compile only when PCI is present.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/acpi/Makefile   | 3 ++-
 drivers/acpi/internal.h | 4 ++++
 2 files changed, 6 insertions(+), 1 deletion(-)
drivers/acpi/Makefile +2 -1
index 7c6afc111d76..bb857421c2e8 100644
@@ -41,7 +41,8 @@ acpi-y				+= ec.o
 acpi-$(CONFIG_ACPI_DOCK)	+= dock.o
 acpi-$(CONFIG_PCI)		+= pci_root.o pci_link.o pci_irq.o
 obj-$(CONFIG_ACPI_MCFG)		+= pci_mcfg.o
-acpi-y				+= acpi_lpss.o acpi_apd.o
+acpi-$(CONFIG_PCI)		+= acpi_lpss.o
+acpi-y				+= acpi_apd.o
 acpi-y				+= acpi_platform.o
 acpi-y				+= acpi_pnp.o
 acpi-$(CONFIG_ARM_AMBA)	+= acpi_amba.o
drivers/acpi/internal.h +4 -0
index 7e6952edb5b0..6a9e1fb8913a 100644
@@ -81,7 +81,11 @@ void acpi_debugfs_init(void);
 #else
 static inline void acpi_debugfs_init(void) { return; }
 #endif
+#ifdef CONFIG_PCI
 void acpi_lpss_init(void);
+#else
+static inline void acpi_lpss_init(void) {}
+#endif
 
 void acpi_apd_init(void);
 
Signature · 1 line
2.19.0
author
#2 lore raw

PATCH v5 2/11 ata: pata_acpi: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. PATA_ACPI is a PCI device driver but the PCI dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/ata/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/ata/Kconfig +1 -1
index 4ca7a6b4eaae..8218db17ebdb 100644
@@ -1091,7 +1091,7 @@ comment "Generic fallback / legacy drivers"
 
 config PATA_ACPI
 	tristate "ACPI firmware driver for PATA"
-	depends on ATA_ACPI && ATA_BMDMA
+	depends on ATA_ACPI && ATA_BMDMA && PCI
 	help
 	  This option enables an ACPI method driver which drives
 	  motherboard PATA controller interfaces through the ACPI
Signature · 1 line
2.19.0
#3 lore raw

Re: [PATCH v5 02/11] ata: pata_acpi: Make PCI dependency explicit

Quoted 18 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. PATA_ACPI is a PCI device driver but the PCI dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/ata/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/ata/Kconfig +0 -0
index 4ca7a6b4eaae..8218db17ebdb 100644

drivers/ata/Kconfig comment "Generic fallback / legacy drivers"

Quoted 9 lines
drivers/ata/Kconfig +1 -2
@@ -1091,7 +1091,7 @@ comment "Generic fallback / legacy drivers"
 
 config PATA_ACPI
        tristate "ACPI firmware driver for PATA"
-       depends on ATA_ACPI && ATA_BMDMA
+       depends on ATA_ACPI && ATA_BMDMA && PCI
        help
          This option enables an ACPI method driver which drives
          motherboard PATA controller interfaces through the ACPI
--

Any objections against this one from anyone?

#4 lore raw

Re: Re: [PATCH v5 02/11] ata: pata_acpi: Make PCI dependency explicit

Quoted 20 lines · Rafael J. Wysocki
Quoted 19 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. PATA_ACPI is a PCI device driver but the PCI dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/ata/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/ata/Kconfig +0 -0
index 4ca7a6b4eaae..8218db17ebdb 100644
@@ -1091,7 +1091,7 @@ comment "Generic fallback / legacy drivers"
 
Quoted 10 lines

drivers/ata/Kconfig comment "Generic fallback / legacy drivers"

Quoted 8 lines
drivers/ata/Kconfig +1 -2
 config PATA_ACPI
        tristate "ACPI firmware driver for PATA"
-       depends on ATA_ACPI && ATA_BMDMA
+       depends on ATA_ACPI && ATA_BMDMA && PCI
        help
          This option enables an ACPI method driver which drives
          motherboard PATA controller interfaces through the ACPI
--

Any objections against this one from anyone?

Patch looks correct to me (pata_acpi is indeed a PCI driver):

but it is best to wait for Jens' ACK if you want to merge it through ACPI tree..

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics

Reviewed-by
#5 lore raw

Re: [PATCH v5 02/11] ata: pata_acpi: Make PCI dependency explicit

Quoted 33 lines · Bartlomiej Zolnierkiewicz
Quoted 20 lines · Rafael J. Wysocki
Quoted 19 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. PATA_ACPI is a PCI device driver but the PCI dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/ata/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/ata/Kconfig +0 -0
index 4ca7a6b4eaae..8218db17ebdb 100644
@@ -1091,7 +1091,7 @@ comment "Generic fallback / legacy drivers"
 
Quoted 10 lines

drivers/ata/Kconfig comment "Generic fallback / legacy drivers"

Quoted 8 lines
drivers/ata/Kconfig +1 -2
 config PATA_ACPI
        tristate "ACPI firmware driver for PATA"
-       depends on ATA_ACPI && ATA_BMDMA
+       depends on ATA_ACPI && ATA_BMDMA && PCI
        help
          This option enables an ACPI method driver which drives
          motherboard PATA controller interfaces through the ACPI
--

Any objections against this one from anyone?

Quoted 6 lines

Patch looks correct to me (pata_acpi is indeed a PCI driver):

but it is best to wait for Jens' ACK if you want to merge it through ACPI tree..

You can add my:

if you want to merge it through the ACPI tree.

Signature · 1 line
Jens Axboe
#6 lore raw

Re: [PATCH v5 02/11] ata: pata_acpi: Make PCI dependency explicit

Quoted 39 lines · Jens Axboe
Quoted 37 lines · Bartlomiej Zolnierkiewicz
Quoted 20 lines · Rafael J. Wysocki
Quoted 19 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. PATA_ACPI is a PCI device driver but the PCI dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/ata/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/ata/Kconfig +0 -0
index 4ca7a6b4eaae..8218db17ebdb 100644
@@ -1091,7 +1091,7 @@ comment "Generic fallback / legacy drivers"
 
Quoted 10 lines

drivers/ata/Kconfig comment "Generic fallback / legacy drivers"

Quoted 8 lines
drivers/ata/Kconfig +1 -2
 config PATA_ACPI
        tristate "ACPI firmware driver for PATA"
-       depends on ATA_ACPI && ATA_BMDMA
+       depends on ATA_ACPI && ATA_BMDMA && PCI
        help
          This option enables an ACPI method driver which drives
          motherboard PATA controller interfaces through the ACPI
--

Any objections against this one from anyone?

Patch looks correct to me (pata_acpi is indeed a PCI driver):

Quoted 8 lines
Quoted 2 lines

but it is best to wait for Jens' ACK if you want to merge it through ACPI tree..

You can add my:

if you want to merge it through the ACPI tree.

I will do that, thanks!

author
#7 lore raw

PATCH v5 3/11 vga-switcheroo: make PCI dependency explicit

In reply to a message not in the archive

This driver depends on the PCI infrastructure but the dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/gpu/vga/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/gpu/vga/Kconfig +1 -0
index b677e5d524e6..d5f1d8e1c6f8 100644
@@ -21,6 +21,7 @@ config VGA_SWITCHEROO
 	bool "Laptop Hybrid Graphics - GPU switching support"
 	depends on X86
 	depends on ACPI
+	depends on PCI
 	select VGA_ARB
 	help
 	  Many laptops released in 2008/9/10 have two GPUs with a multiplexer
Signature · 6 lines
2.19.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author
#8 lore raw

PATCH v5 4/11 platform/x86: intel_ips: make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. Ipss driver is a PCI device driver but this has not been mentioned anywhere in Kconfig.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/platform/x86/Kconfig +1 -1
index e3b62c2ee8d1..b36ea14b41ad 100644
@@ -1009,7 +1009,7 @@ config INTEL_MFLD_THERMAL
 
 config INTEL_IPS
 	tristate "Intel Intelligent Power Sharing"
-	depends on ACPI
+	depends on ACPI && PCI
 	---help---
 	  Intel Calpella platforms support dynamic power sharing between the
 	  CPU and GPU, maximizing performance in a given TDP.  This driver,
Signature · 1 line
2.19.0
Acked-by on v5 4/11
#9 lore raw

Re: [PATCH v5 04/11] platform/x86: intel_ips: make PCI dependency explicit

Quoted 7 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. Ipss driver is a PCI device driver but this has not been mentioned anywhere in Kconfig.

Quoted 22 lines
Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/platform/x86/Kconfig +1 -2
index e3b62c2ee8d1..b36ea14b41ad 100644
@@ -1009,7 +1009,7 @@ config INTEL_MFLD_THERMAL
 
 config INTEL_IPS
        tristate "Intel Intelligent Power Sharing"
-       depends on ACPI
+       depends on ACPI && PCI
        ---help---
          Intel Calpella platforms support dynamic power sharing between the
          CPU and GPU, maximizing performance in a given TDP.  This driver,
--

2.19.0

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#10 lore raw

PATCH v5 5/11 platform/x86: intel_pmc: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. Code relies on PCI for execution. Specify this in the Kconfig.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/platform/x86/Kconfig +1 -1
index b36ea14b41ad..7afb96cb1cd6 100644
@@ -1174,7 +1174,7 @@ config INTEL_SMARTCONNECT
 
 config INTEL_PMC_IPC
 	tristate "Intel PMC IPC Driver"
-	depends on ACPI
+	depends on ACPI && PCI
 	---help---
 	This driver provides support for PMC control on some Intel platforms.
 	The PMC is an ARC processor which defines IPC commands for communication
Signature · 1 line
2.19.0
#11 lore raw

Re: [PATCH v5 05/11] platform/x86: intel_pmc: Make PCI dependency explicit

Quoted 7 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. Code relies on PCI for execution. Specify this in the Kconfig.

Unfortunately, this is not how it should be fixed.

First of all, we (Heikki, Sathya, me — all in Cc list) that this driver should be refactored and unified with intel_scu_ipc. Then, it has two drivers inside and PCI is just a glue for it which is optional. Thus, needs to be split accordingly.

Quoted 22 lines
Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/platform/x86/Kconfig +1 -2
index b36ea14b41ad..7afb96cb1cd6 100644
@@ -1174,7 +1174,7 @@ config INTEL_SMARTCONNECT
 
 config INTEL_PMC_IPC
        tristate "Intel PMC IPC Driver"
-       depends on ACPI
+       depends on ACPI && PCI
        ---help---
        This driver provides support for PMC control on some Intel platforms.
        The PMC is an ARC processor which defines IPC commands for communication
--

2.19.0

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#12 lore raw

Re: [PATCH v5 05/11] platform/x86: intel_pmc: Make PCI dependency explicit

Quoted 6 lines · Andy Shevchenko

Unfortunately, this is not how it should be fixed.

First of all, we (Heikki, Sathya, me — all in Cc list) that this driver should be refactored and unified with intel_scu_ipc. Then, it has two drivers inside and PCI is just a glue for it which is optional. Thus, needs to be split accordingly.

OK. Two questions:

1. How should I change this now?
2. Can you change the Kconfig accordingly after you merge the two drivers?

I'm focused on the build failure at this moment.

author
#13 lore raw

PATCH v5 6/11 platform/x86: apple-gmux: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver depends on the PCI infrastructure but the dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/platform/x86/Kconfig +1 -0
index 7afb96cb1cd6..89f4b86244a7 100644
@@ -1136,6 +1136,7 @@ config SAMSUNG_Q10
 config APPLE_GMUX
 	tristate "Apple Gmux Driver"
 	depends on ACPI
+	depends on PCI
 	depends on PNP
 	depends on BACKLIGHT_CLASS_DEVICE
 	depends on BACKLIGHT_APPLE=n || BACKLIGHT_APPLE
Signature · 1 line
2.19.0
Reviewed-by on v5 6/11
#14 lore raw

Re: [PATCH v5 06/11] platform/x86: apple-gmux: Make PCI dependency explicit

Quoted 8 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver depends on the PCI infrastructure but the dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
Quoted 18 lines
---
 drivers/platform/x86/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/platform/x86/Kconfig +1 -0
index 7afb96cb1cd6..89f4b86244a7 100644
@@ -1136,6 +1136,7 @@ config SAMSUNG_Q10
 config APPLE_GMUX
 	tristate "Apple Gmux Driver"
 	depends on ACPI
+	depends on PCI
 	depends on PNP
 	depends on BACKLIGHT_CLASS_DEVICE
 	depends on BACKLIGHT_APPLE=n || BACKLIGHT_APPLE
Signature · 1 line
2.19.0
#15 lore raw

Re: [PATCH v5 06/11] platform/x86: apple-gmux: Make PCI dependency explicit

In reply to Sinan Kaya
Quoted 18 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver depends on the PCI infrastructure but the dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/platform/x86/Kconfig +0 -0
index 7afb96cb1cd6..89f4b86244a7 100644

drivers/platform/x86/Kconfig config SAMSUNG_Q10 · line 1140

Quoted 5 lines
drivers/platform/x86/Kconfig +1 -0
@@ -1136,6 +1136,7 @@ config SAMSUNG_Q10
 config APPLE_GMUX
        tristate "Apple Gmux Driver"
 
        depends on ACPI
+       depends on PCI

Can you stick with the same pattern you have used for the rest in this series? I.e.

         depends on ACPI && PCI
Quoted 6 lines
drivers/platform/x86/Kconfig +0 -0
        depends on PNP
        depends on BACKLIGHT_CLASS_DEVICE
        depends on BACKLIGHT_APPLE=n || BACKLIGHT_APPLE

--
2.19.0

Signature · 2 lines
With Best Regards,
Andy Shevchenko
Reviewed-by Acked-by
#16 lore raw

Re: [PATCH v5 06/11] platform/x86: apple-gmux: Make PCI dependency explicit

Quoted 23 lines · Andy Shevchenko
Quoted 20 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver depends on the PCI infrastructure but the dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/platform/x86/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/platform/x86/Kconfig +0 -0
index 7afb96cb1cd6..89f4b86244a7 100644
@@ -1136,6 +1136,7 @@ config SAMSUNG_Q10
 config APPLE_GMUX
        tristate "Apple Gmux Driver"
Quoted 6 lines

drivers/platform/x86/Kconfig config SAMSUNG_Q10 · line 1139

Quoted 2 lines
drivers/platform/x86/Kconfig +1 -0
        depends on ACPI
+       depends on PCI

Can you stick with the same pattern you have used for the rest in this series? I.e.

         depends on ACPI && PCI

After satisfying it

Quoted 13 lines
Quoted 8 lines
drivers/platform/x86/Kconfig +0 -1
        depends on PNP
        depends on BACKLIGHT_CLASS_DEVICE
        depends on BACKLIGHT_APPLE=n || BACKLIGHT_APPLE
--

2.19.0

--
With Best Regards,
Andy Shevchenko

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#17 lore raw

PATCH v5 7/11 drivers: thermal: int3406_thermal: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. Need CONFIG_PCI to be set in order to be able to use this driver.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/thermal/intel/int340x_thermal/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/thermal/intel/int340x_thermal/Kconfig +1 -1
index 0582bd12a239..fba1976d5f8d 100644
@@ -31,7 +31,7 @@ if INT340X_THERMAL
 
 config INT3406_THERMAL
 	tristate "ACPI INT3406 display thermal driver"
-	depends on ACPI_VIDEO
+	depends on ACPI_VIDEO && PCI
 	help
 	  The display thermal device represents the LED/LCD display panel
 	  that may or may not include touch support. The main function of
Signature · 1 line
2.19.0
author
#18 lore raw

PATCH v5 8/11 ASoC: Intel: atom: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This code relies on IOSF_MBI and IOSF_MBI depends on PCI. For this reason, add a direct dependency on CONFIG_PCI to the IOSF_MBI driver.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 sound/soc/intel/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/soc/intel/Kconfig +1 -1
index 2fd1b61e8331..b0764b2fe001 100644
@@ -91,7 +91,7 @@ config SND_SST_ATOM_HIFI2_PLATFORM_PCI
 config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
 	tristate "ACPI HiFi2 (Baytrail, Cherrytrail) Platforms"
 	default ACPI
-	depends on X86 && ACPI
+	depends on X86 && ACPI && PCI
 	select SND_SST_IPC_ACPI
 	select SND_SST_ATOM_HIFI2_PLATFORM
 	select SND_SOC_ACPI_INTEL_MATCH
Signature · 1 line
2.19.0
#19 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

I have three opens with this ACPI/PCI change

1. the baseline change fails on my cross-compilation checks, see below the result of the attached script (simplification of the one I use to avoid 0day reports).

2. there are different patterns to express the dependency on PCI e.g.

 config MMC_SDHCI_ACPI
     tristate "SDHCI support for ACPI enumerated SDHCI controllers" depends on MMC_SDHCI && ACPI
-    select IOSF_MBI if X86
+    select IOSF_MBI if (X86 && PCI)

but

config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
     tristate "ACPI HiFi2 (Baytrail, Cherrytrail) Platforms" default ACPI
-    depends on X86 && ACPI
+    depends on X86 && ACPI && PCI
     select SND_SST_IPC_ACPI
     select SND_SST_ATOM_HIFI2_PLATFORM
     select SND_SOC_ACPI_INTEL_MATCH

IOSF is only needed for Baytrail-CR detection, and the code will compile fine without it, so maybe it'd be a better model if you used the following diff?

sound/soc/intel/Kconfig +0 -0
index 2fd1b61e8331..68af0ea5c96c 100644
@@ -95,7 +95,7 @@ config SND_SST_ATOM_HIFI2_PLATFORM_ACPI

        select SND_SST_IPC_ACPI
        select SND_SST_ATOM_HIFI2_PLATFORM
        select SND_SOC_ACPI_INTEL_MATCH
-       select IOSF_MBI
+       select IOSF_MBI if PCI

3. All the Intel machine drivers depend on X86_INTEL_LPSS which depends on PCI. But for Baytrail/Haswell/Broadwell we have only a dependency on ACPI, so we expose drivers that can be selected but fail on probe since there are no machine drivers. I am not sure if we want to be strict and only expose meaningful configurations, or allow for more compilations tests and corner cases?

-Pierre

cross-compilation issue:

git checkout next-20190102

make CROSS_COMPILE=/opt/gcc-8.1.0-nolibc/ia64-linux/bin/ia64-linux- --jobs=16 allmodconfig ARCH=ia64 HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  YACC    scripts/kconfig/zconf.tab.c
  LEX     scripts/kconfig/zconf.lex.c
  HOSTCC  scripts/kconfig/confdata.o
  HOSTCC  scripts/kconfig/expr.o
  HOSTCC  scripts/kconfig/symbol.o
  YACC    scripts/kconfig/zconf.tab.h
  HOSTCC  scripts/kconfig/preprocess.o
  HOSTCC  scripts/kconfig/zconf.tab.o
  HOSTCC  scripts/kconfig/zconf.lex.o
  HOSTLD  scripts/kconfig/conf
scripts/kconfig/conf  --allmodconfig Kconfig
arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128:    choice <choice> contains symbol IA64_HP_SIM arch/ia64/Kconfig:202:    symbol IA64_HP_SIM is part of choice PM kernel/power/Kconfig:144:    symbol PM is selected by PM_SLEEP kernel/power/Kconfig:104:    symbol PM_SLEEP depends on HIBERNATE_CALLBACKS kernel/power/Kconfig:31:    symbol HIBERNATE_CALLBACKS is selected by HIBERNATION kernel/power/Kconfig:34:    symbol HIBERNATION depends on SWAP
init/Kconfig:250:    symbol SWAP depends on BLOCK
block/Kconfig:5:    symbol BLOCK is selected by UBIFS_FS
fs/ubifs/Kconfig:1:    symbol UBIFS_FS depends on MISC_FILESYSTEMS
fs/Kconfig:227:    symbol MISC_FILESYSTEMS is selected by ACPI_APEI drivers/acpi/apei/Kconfig:8:    symbol ACPI_APEI depends on ACPI drivers/acpi/Kconfig:9:    symbol ACPI depends on ARCH_SUPPORTS_ACPI drivers/acpi/Kconfig:6:    symbol ARCH_SUPPORTS_ACPI is selected by IA64_HP_SIM arch/ia64/Kconfig:202:    symbol IA64_HP_SIM is part of choice <choice>
For a resolution refer to Documentation/kbuild/kconfig-language.txt
subsection "Kconfig recursive dependency limitations"

#!/bin/bash

set -e

test_fast() {

PLATFORM="$1"
make mrproper
../make.cross allmodconfig ARCH=$PLATFORM
../make.cross olddefconfig ARCH=$PLATFORM
../make.cross ARCH=$PLATFORM modules_prepare
}

test_compile() {

PLATFORM="$1"
make mrproper

make defconfig ARCH=$PLATFORM

echo "modules first"
../make.cross ARCH=$PLATFORM
if [ "$PLATFORM" = "x86_64" ]; then
../make.cross bindeb-pkg ARCH=$PLATFORM
fi

echo "all yes second"
perl -pi.bak -e 's/=m/=y/g' .config
../make.cross olddefconfig ARCH=$PLATFORM
../make.cross ARCH=$PLATFORM
if [ "$PLATFORM" = "x86_64" ]; then
../make.cross bindeb-pkg ARCH=$PLATFORM
fi

}

test_platform() {

    PLATFORM="$1"
       
    test_fast    $PLATFORM

# test_compile $PLATFORM
}

test_platform x86_64
test_platform i386
test_platform ia64
test_platform arm
test_platform arm64
test_platform sh
test_platform mips
test_platform s390
test_platform openrisc
test_platform sparc
test_platform sparc64
test_platform m68k

echo "X-compilation check PASS"

#20 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 6 lines · Pierre-Louis Bossart

I have three opens with this ACPI/PCI change

1. the baseline change fails on my cross-compilation checks, see below the result of the attached script (simplification of the one I use to avoid 0day reports).

What baseline change?

That failure is not related to PCI if I'm not missing anything.

Quoted 28 lines

2. there are different patterns to express the dependency on PCI e.g.

config MMC_SDHCI_ACPI
tristate "SDHCI support for ACPI enumerated SDHCI controllers" depends on MMC_SDHCI && ACPI

(quoted diff) +1 -1
-    select IOSF_MBI if X86
+    select IOSF_MBI if (X86 && PCI)

but

config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
tristate "ACPI HiFi2 (Baytrail, Cherrytrail) Platforms" default ACPI

(quoted diff) +1 -1
-    depends on X86 && ACPI
+    depends on X86 && ACPI && PCI
      select SND_SST_IPC_ACPI
      select SND_SST_ATOM_HIFI2_PLATFORM
      select SND_SOC_ACPI_INTEL_MATCH

IOSF is only needed for Baytrail-CR detection, and the code will compile fine without it, so maybe it'd be a better model if you used the following diff?

sound/soc/intel/Kconfig +0 -0
index 2fd1b61e8331..68af0ea5c96c 100644

sound/soc/intel/Kconfig config SND_SST_ATOM_HIFI2_PLATFORM_ACPI · line 98

Quoted 5 lines
sound/soc/intel/Kconfig +1 -1
@@ -95,7 +95,7 @@ config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
         select SND_SST_IPC_ACPI
         select SND_SST_ATOM_HIFI2_PLATFORM
         select SND_SOC_ACPI_INTEL_MATCH
-       select IOSF_MBI
+       select IOSF_MBI if PCI

Well, does it actually make sense to ever set
SND_SST_ATOM_HIFI2_PLATFORM_ACPI without PCI?

Quoted 6 lines

3. All the Intel machine drivers depend on X86_INTEL_LPSS which depends on PCI. But for Baytrail/Haswell/Broadwell we have only a dependency on ACPI, so we expose drivers that can be selected but fail on probe since there are no machine drivers. I am not sure if we want to be strict and only expose meaningful configurations, or allow for more compilations tests and corner cases?

I would only expose meaningful configurations to start with and then *maybe* relax that going forward as long as the benefit is worth it.

Cheers,
Rafael

author
#21 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

In reply to Pierre-Louis Bossart
Quoted 7 lines · Pierre-Louis Bossart

I have three opens with this ACPI/PCI change

1. the baseline change fails on my cross-compilation checks, see below the result of the attached script (simplification of the one I use to avoid 0day reports).

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Quoted 14 lines

2. there are different patterns to express the dependency on PCI e.g.

config MMC_SDHCI_ACPI
tristate "SDHCI support for ACPI enumerated SDHCI controllers" depends on MMC_SDHCI && ACPI

(quoted diff) +1 -1
-    select IOSF_MBI if X86
+    select IOSF_MBI if (X86 && PCI)

but

config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
tristate "ACPI HiFi2 (Baytrail, Cherrytrail) Platforms" default ACPI

Review

Quoted 5 lines
(quoted diff) +1 -1
-    depends on X86 && ACPI
+    depends on X86 && ACPI && PCI
      select SND_SST_IPC_ACPI
      select SND_SST_ATOM_HIFI2_PLATFORM
      select SND_SOC_ACPI_INTEL_MATCH

I matched depends line to

depends on X86 && ACPI && PCI

for MMC_SDHCI_ACPI per feedback from Rafael on V5. This should resolve the inconsistency.

Quoted 15 lines

IOSF is only needed for Baytrail-CR detection, and the code will compile fine without it, so maybe it'd be a better model if you used the following diff?

sound/soc/intel/Kconfig +1 -1
index 2fd1b61e8331..68af0ea5c96c 100644
@@ -95,7 +95,7 @@ config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
         select SND_SST_IPC_ACPI
         select SND_SST_ATOM_HIFI2_PLATFORM
         select SND_SOC_ACPI_INTEL_MATCH
-       select IOSF_MBI
+       select IOSF_MBI if PCI
 
Quoted 6 lines

3. All the Intel machine drivers depend on X86_INTEL_LPSS which depends on PCI. But for Baytrail/Haswell/Broadwell we have only a dependency on ACPI, so we expose drivers that can be selected but fail on probe since there are no machine drivers. I am not sure if we want to be strict and only expose meaningful configurations, or allow for more compilations tests and corner cases?

Hopefully, v5 resolves this too with

depends on X86 && ACPI && PCI

Let me know otherwise.

#22 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 11 lines

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Are you saying there's a newer series that fixes this issue for both allyesconfig and allmodconfig?

if yes, then we're good.

Quoted 20 lines
Quoted 15 lines

2. there are different patterns to express the dependency on PCI e.g.

config MMC_SDHCI_ACPI
tristate "SDHCI support for ACPI enumerated SDHCI controllers" depends on MMC_SDHCI && ACPI

(quoted diff) +1 -1
-    select IOSF_MBI if X86
+    select IOSF_MBI if (X86 && PCI)

but

config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
tristate "ACPI HiFi2 (Baytrail, Cherrytrail) Platforms" default ACPI

Review

Quoted 5 lines
(quoted diff) +1 -1
-    depends on X86 && ACPI
+    depends on X86 && ACPI && PCI
       select SND_SST_IPC_ACPI
       select SND_SST_ATOM_HIFI2_PLATFORM
       select SND_SOC_ACPI_INTEL_MATCH
Quoted 6 lines

I matched depends line to

depends on X86 && ACPI && PCI

for MMC_SDHCI_ACPI per feedback from Rafael on V5. This should resolve the inconsistency.

ok, I didn't see the delta

Quoted 25 lines
Quoted 17 lines

IOSF is only needed for Baytrail-CR detection, and the code will compile fine without it, so maybe it'd be a better model if you used the following diff?

sound/soc/intel/Kconfig +1 -1
index 2fd1b61e8331..68af0ea5c96c 100644
@@ -95,7 +95,7 @@ config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
          select SND_SST_IPC_ACPI
          select SND_SST_ATOM_HIFI2_PLATFORM
          select SND_SOC_ACPI_INTEL_MATCH
-       select IOSF_MBI
+       select IOSF_MBI if PCI
 
Quoted 6 lines

3. All the Intel machine drivers depend on X86_INTEL_LPSS which depends on PCI. But for Baytrail/Haswell/Broadwell we have only a dependency on ACPI, so we expose drivers that can be selected but fail on probe since there are no machine drivers. I am not sure if we want to be strict and only expose meaningful configurations, or allow for more compilations tests and corner cases?

Hopefully, v5 resolves this too with

Quoted 3 lines

depends on X86 && ACPI && PCI

Let me know otherwise.

it doesn't but that's not a good enough reason to lay on the tracks. I'll follow-up with a cleanup for the Intel audio parts when this series is merged. The PCI dependency could be moved to the top-level since it's pretty much required for all platforms except for compilation tests, and there are multiple dependencies that repeated for no good reason, so FWIW

author
#23 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 10 lines · Pierre-Louis Bossart
Quoted 10 lines

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

Quoted 7 lines
Quoted 2 lines

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Are you saying there's a newer series that fixes this issue for both allyesconfig and allmodconfig?

if yes, then we're good.

No, I haven't fixed ia64 kconfig issue. That's somebody else's job. I used allyesconfig to find out all compilation issues on x86 arch to come up with this patchset.

Quoted 65 lines
Quoted 21 lines
Quoted 16 lines

2. there are different patterns to express the dependency on PCI e.g.

config MMC_SDHCI_ACPI
tristate "SDHCI support for ACPI enumerated SDHCI controllers" depends on MMC_SDHCI && ACPI

(quoted diff) +1 -1
-    select IOSF_MBI if X86
+    select IOSF_MBI if (X86 && PCI)

but

config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
tristate "ACPI HiFi2 (Baytrail, Cherrytrail) Platforms" default ACPI

Review

Quoted 5 lines
(quoted diff) +1 -1
-    depends on X86 && ACPI
+    depends on X86 && ACPI && PCI
       select SND_SST_IPC_ACPI
       select SND_SST_ATOM_HIFI2_PLATFORM
       select SND_SOC_ACPI_INTEL_MATCH
Quoted 6 lines

I matched depends line to

depends on X86 && ACPI && PCI

for MMC_SDHCI_ACPI per feedback from Rafael on V5. This should resolve the inconsistency.

ok, I didn't see the delta

Quoted 25 lines
Quoted 17 lines

IOSF is only needed for Baytrail-CR detection, and the code will compile fine without it, so maybe it'd be a better model if you used the following diff?

sound/soc/intel/Kconfig +1 -1
index 2fd1b61e8331..68af0ea5c96c 100644
@@ -95,7 +95,7 @@ config SND_SST_ATOM_HIFI2_PLATFORM_ACPI
          select SND_SST_IPC_ACPI
          select SND_SST_ATOM_HIFI2_PLATFORM
          select SND_SOC_ACPI_INTEL_MATCH
-       select IOSF_MBI
+       select IOSF_MBI if PCI
 
Quoted 6 lines

3. All the Intel machine drivers depend on X86_INTEL_LPSS which depends on PCI. But for Baytrail/Haswell/Broadwell we have only a dependency on ACPI, so we expose drivers that can be selected but fail on probe since there are no machine drivers. I am not sure if we want to be strict and only expose meaningful configurations, or allow for more compilations tests and corner cases?

Hopefully, v5 resolves this too with

Quoted 3 lines

depends on X86 && ACPI && PCI

Let me know otherwise.

it doesn't but that's not a good enough reason to lay on the tracks. I'll follow-up with a cleanup for the Intel audio parts when this series is merged. The PCI dependency could be moved to the top-level since it's pretty much required for all platforms except for compilation tests, and there are multiple dependencies that repeated for no good reason, so FWIW

#24 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 13 lines · Sinan Kaya
Quoted 11 lines · Pierre-Louis Bossart
Quoted 11 lines

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Quoted 8 lines
Quoted 4 lines

Are you saying there's a newer series that fixes this issue for both allyesconfig and allmodconfig?

if yes, then we're good.

No, I haven't fixed ia64 kconfig issue. That's somebody else's job. I used allyesconfig to find out all compilation issues on x86 arch to come up with this patchset.

Nothing makes me cringe more than "somebody else's job" statements. In this case, there is obviously a correlation with your ACPI changes since the circular dependency happens because of the ACPI symbol.

arch/ia64/Kconfig:126:error: recursive dependency detected! arch/ia64/Kconfig:126:    choice <choice> contains symbol IA64_HP_SIM arch/ia64/Kconfig:200:    symbol IA64_HP_SIM is part of choice PM kernel/power/Kconfig:144:    symbol PM is selected by PM_SLEEP kernel/power/Kconfig:104:    symbol PM_SLEEP depends on HIBERNATE_CALLBACKS kernel/power/Kconfig:31:    symbol HIBERNATE_CALLBACKS is selected by HIBERNATION kernel/power/Kconfig:34:    symbol HIBERNATION depends on SWAP init/Kconfig:250:    symbol SWAP depends on BLOCK
block/Kconfig:5:    symbol BLOCK is selected by UBIFS_FS fs/ubifs/Kconfig:1:    symbol UBIFS_FS depends on MISC_FILESYSTEMS fs/Kconfig:220:    symbol MISC_FILESYSTEMS is selected by ACPI_APEI drivers/acpi/apei/Kconfig:8:    symbol ACPI_APEI depends on ACPI drivers/acpi/Kconfig:9:    symbol ACPI depends on ARCH_SUPPORTS_ACPI <<<< LOOK HERE drivers/acpi/Kconfig:6:    symbol ARCH_SUPPORTS_ACPI is selected by IA64_HP_SIM arch/ia64/Kconfig:200:    symbol IA64_HP_SIM is part of choice <choice>

At any rate, a 3 mn git bisect tells me the circular dependency is exposed by this change:

f3fd6cd74fedf99b6060f75df00943fda13b65f2 is the first bad commit commit f3fd6cd74fedf99b6060f75df00943fda13b65f2
Author: Chandan Rajendra <[email protected]> Date:   Sat Dec 8 12:21:38 2018 +0530

    fscrypt: remove filesystem specific build config option

    In order to have a common code base for fscrypt "post read" processing for all filesystems which support encryption, this commit removes filesystem specific build config option (e.g. CONFIG_EXT4_FS_ENCRYPTION) and replaces it with a build option (i.e. CONFIG_FS_ENCRYPTION) whose value affects all the filesystems making use of fscrypt.

-Pierre

author
#25 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

+Tony

Quoted 21 lines · Pierre-Louis Bossart
Quoted 18 lines · Sinan Kaya
Quoted 15 lines · Pierre-Louis Bossart
Quoted 11 lines

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Are you saying there's a newer series that fixes this issue for both allyesconfig and allmodconfig?

if yes, then we're good.

Quoted 8 lines
Quoted 3 lines

No, I haven't fixed ia64 kconfig issue. That's somebody else's job. I used allyesconfig to find out all compilation issues on x86 arch to come up with this patchset.

Nothing makes me cringe more than "somebody else's job" statements. In this case, there is obviously a correlation with your ACPI changes since the circular dependency happens because of the ACPI symbol.

I agree that it needs to be fixed. I am not convinced that the issue is related to my change.
Your log is pointing to an inconsistency in ia64 kconfig.

Quoted 7 lines

arch/ia64/Kconfig:126:error: recursive dependency detected!
arch/ia64/Kconfig:126: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice PM
kernel/power/Kconfig:144: symbol PM is selected by PM_SLEEP
kernel/power/Kconfig:104: symbol PM_SLEEP depends on HIBERNATE_CALLBACKS
kernel/power/Kconfig:31: symbol HIBERNATE_CALLBACKS is selected by
HIBERNATION

Quoted 8 lines

kernel/power/Kconfig:34: symbol HIBERNATION depends on SWAP
init/Kconfig:250: symbol SWAP depends on BLOCK
block/Kconfig:5: symbol BLOCK is selected by UBIFS_FS
fs/ubifs/Kconfig:1: symbol UBIFS_FS depends on MISC_FILESYSTEMS
fs/Kconfig:220: symbol MISC_FILESYSTEMS is selected by ACPI_APEI
drivers/acpi/apei/Kconfig:8: symbol ACPI_APEI depends on ACPI
drivers/acpi/Kconfig:9: symbol ACPI depends on ARCH_SUPPORTS_ACPI
<<<< LOOK HERE

I am still having hard time seeing how this issue is exposed by removing PCI dependency from ACPi.

Rafael, Tony:

can you help me?

Quoted 27 lines

drivers/acpi/Kconfig:6: symbol ARCH_SUPPORTS_ACPI is selected by
IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice <choice>

At any rate, a 3 mn git bisect tells me the circular dependency is exposed by this change:

f3fd6cd74fedf99b6060f75df00943fda13b65f2 is the first bad commit commit f3fd6cd74fedf99b6060f75df00943fda13b65f2

Author: Chandan Rajendra <[email protected]>
Date:   Sat Dec 8 12:21:38 2018 +0530

fscrypt: remove filesystem specific build config option

In order to have a common code base for fscrypt "post read" processing for all filesystems which support encryption, this commit removes filesystem specific build config option (e.g.
CONFIG_EXT4_FS_ENCRYPTION)
and replaces it with a build option (i.e. CONFIG_FS_ENCRYPTION) whose value affects all the filesystems making use of fscrypt.

-Pierre

#26 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

In reply to Pierre-Louis Bossart
Quoted 64 lines · IST Pierre-Louis Bossart
Quoted 13 lines · Sinan Kaya
Quoted 11 lines · Pierre-Louis Bossart
Quoted 11 lines

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Quoted 8 lines
Quoted 4 lines

Are you saying there's a newer series that fixes this issue for both allyesconfig and allmodconfig?

if yes, then we're good.

No, I haven't fixed ia64 kconfig issue. That's somebody else's job. I used allyesconfig to find out all compilation issues on x86 arch to come up with this patchset.

Nothing makes me cringe more than "somebody else's job" statements. In this case, there is obviously a correlation with your ACPI changes since the circular dependency happens because of the ACPI symbol.

arch/ia64/Kconfig:126:error: recursive dependency detected!
arch/ia64/Kconfig:126: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice PM
kernel/power/Kconfig:144: symbol PM is selected by PM_SLEEP
kernel/power/Kconfig:104: symbol PM_SLEEP depends on HIBERNATE_CALLBACKS
kernel/power/Kconfig:31: symbol HIBERNATE_CALLBACKS is selected by
HIBERNATION
kernel/power/Kconfig:34: symbol HIBERNATION depends on SWAP
init/Kconfig:250: symbol SWAP depends on BLOCK
block/Kconfig:5: symbol BLOCK is selected by UBIFS_FS
fs/ubifs/Kconfig:1: symbol UBIFS_FS depends on MISC_FILESYSTEMS
fs/Kconfig:220: symbol MISC_FILESYSTEMS is selected by ACPI_APEI
drivers/acpi/apei/Kconfig:8: symbol ACPI_APEI depends on ACPI
drivers/acpi/Kconfig:9: symbol ACPI depends on ARCH_SUPPORTS_ACPI
<<<< LOOK HERE
drivers/acpi/Kconfig:6: symbol ARCH_SUPPORTS_ACPI is selected by
IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice <choice>

At any rate, a 3 mn git bisect tells me the circular dependency is exposed by this change:

f3fd6cd74fedf99b6060f75df00943fda13b65f2 is the first bad commit commit f3fd6cd74fedf99b6060f75df00943fda13b65f2

Author: Chandan Rajendra <[email protected]>
Date:   Sat Dec 8 12:21:38 2018 +0530

fscrypt: remove filesystem specific build config option

In order to have a common code base for fscrypt "post read" processing for all filesystems which support encryption, this commit removes filesystem specific build config option (e.g.
CONFIG_EXT4_FS_ENCRYPTION)
and replaces it with a build option (i.e. CONFIG_FS_ENCRYPTION) whose value affects all the filesystems making use of fscrypt.

Quoted 3 lines

FWIW, The patch at https://patchwork.kernel.org/patch/10725883/ fixes this problem by removing "select BLOCK if FS_ENCRYPTION" from fs/ubifs/Kconfig.

Signature · 1 line
chandan
#27 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 66 lines · Chandan Rajendra
Quoted 64 lines · IST Pierre-Louis Bossart
Quoted 13 lines · Sinan Kaya
Quoted 11 lines · Pierre-Louis Bossart
Quoted 11 lines

This is pointing to a kconfig issue on ia64 arch.

arch/ia64/Kconfig:128:error: recursive dependency detected!
arch/ia64/Kconfig:128: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:202: symbol IA64_HP_SIM is part of choice PM

IA64_HP_SIM is both a choice and is selected.

I did allyesconfig and disabled PCI afterwards to find all the issues on this patchset.

Quoted 8 lines
Quoted 4 lines

Are you saying there's a newer series that fixes this issue for both allyesconfig and allmodconfig?

if yes, then we're good.

No, I haven't fixed ia64 kconfig issue. That's somebody else's job. I used allyesconfig to find out all compilation issues on x86 arch to come up with this patchset.

Nothing makes me cringe more than "somebody else's job" statements. In this case, there is obviously a correlation with your ACPI changes since the circular dependency happens because of the ACPI symbol.

arch/ia64/Kconfig:126:error: recursive dependency detected!
arch/ia64/Kconfig:126: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice PM
kernel/power/Kconfig:144: symbol PM is selected by PM_SLEEP
kernel/power/Kconfig:104: symbol PM_SLEEP depends on HIBERNATE_CALLBACKS
kernel/power/Kconfig:31: symbol HIBERNATE_CALLBACKS is selected by
HIBERNATION
kernel/power/Kconfig:34: symbol HIBERNATION depends on SWAP
init/Kconfig:250: symbol SWAP depends on BLOCK
block/Kconfig:5: symbol BLOCK is selected by UBIFS_FS
fs/ubifs/Kconfig:1: symbol UBIFS_FS depends on MISC_FILESYSTEMS
fs/Kconfig:220: symbol MISC_FILESYSTEMS is selected by ACPI_APEI
drivers/acpi/apei/Kconfig:8: symbol ACPI_APEI depends on ACPI
drivers/acpi/Kconfig:9: symbol ACPI depends on ARCH_SUPPORTS_ACPI
<<<< LOOK HERE
drivers/acpi/Kconfig:6: symbol ARCH_SUPPORTS_ACPI is selected by
IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice <choice>

At any rate, a 3 mn git bisect tells me the circular dependency is exposed by this change:

f3fd6cd74fedf99b6060f75df00943fda13b65f2 is the first bad commit commit f3fd6cd74fedf99b6060f75df00943fda13b65f2

Author: Chandan Rajendra <[email protected]>
Date:   Sat Dec 8 12:21:38 2018 +0530

fscrypt: remove filesystem specific build config option

In order to have a common code base for fscrypt "post read" processing for all filesystems which support encryption, this commit removes filesystem specific build config option (e.g.
CONFIG_EXT4_FS_ENCRYPTION)
and replaces it with a build option (i.e. CONFIG_FS_ENCRYPTION) whose value affects all the filesystems making use of fscrypt.

Quoted 6 lines
Quoted 3 lines

FWIW, The patch at https://patchwork.kernel.org/patch/10725883/ fixes this problem by removing "select BLOCK if FS_ENCRYPTION" from fs/ubifs/Kconfig.

OK

Pierre-Louis, can you check if this patch makes the issue go away for you, please?

#28 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 40 lines
Quoted 40 lines
Quoted 40 lines

arch/ia64/Kconfig:126:error: recursive dependency detected!
arch/ia64/Kconfig:126: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice PM
kernel/power/Kconfig:144: symbol PM is selected by PM_SLEEP
kernel/power/Kconfig:104: symbol PM_SLEEP depends on HIBERNATE_CALLBACKS
kernel/power/Kconfig:31: symbol HIBERNATE_CALLBACKS is selected by
HIBERNATION
kernel/power/Kconfig:34: symbol HIBERNATION depends on SWAP
init/Kconfig:250: symbol SWAP depends on BLOCK
block/Kconfig:5: symbol BLOCK is selected by UBIFS_FS
fs/ubifs/Kconfig:1: symbol UBIFS_FS depends on MISC_FILESYSTEMS
fs/Kconfig:220: symbol MISC_FILESYSTEMS is selected by ACPI_APEI
drivers/acpi/apei/Kconfig:8: symbol ACPI_APEI depends on ACPI
drivers/acpi/Kconfig:9: symbol ACPI depends on ARCH_SUPPORTS_ACPI
<<<< LOOK HERE
drivers/acpi/Kconfig:6: symbol ARCH_SUPPORTS_ACPI is selected by
IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice <choice>

At any rate, a 3 mn git bisect tells me the circular dependency is exposed by this change:

f3fd6cd74fedf99b6060f75df00943fda13b65f2 is the first bad commit commit f3fd6cd74fedf99b6060f75df00943fda13b65f2

Author: Chandan Rajendra <[email protected]>
Date:   Sat Dec 8 12:21:38 2018 +0530

fscrypt: remove filesystem specific build config option

In order to have a common code base for fscrypt "post read" processing for all filesystems which support encryption, this commit removes filesystem specific build config option (e.g.
CONFIG_EXT4_FS_ENCRYPTION)
and replaces it with a build option (i.e. CONFIG_FS_ENCRYPTION) whose value affects all the filesystems making use of fscrypt.

Quoted 6 lines
Quoted 2 lines

FWIW, The patch at https://patchwork.kernel.org/patch/10725883/ fixes this problem by removing "select BLOCK if FS_ENCRYPTION" from fs/ubifs/Kconfig.

OK

Pierre-Louis, can you check if this patch makes the issue go away for you, please?

Wondering if Chandan provided the right pointer, I wasn't able to apply this patch, but commenting out "select BLOCK if FS_ENCRYPTION" in fs/ubifs/Kconfig makes the circular dependency go away. All good for me.

#29 lore raw

Re: [alsa-devel] [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 42 lines · IST Pierre-Louis Bossart
Quoted 42 lines
Quoted 42 lines
Quoted 37 lines

arch/ia64/Kconfig:126:error: recursive dependency detected!
arch/ia64/Kconfig:126: choice <choice> contains symbol IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice PM
kernel/power/Kconfig:144: symbol PM is selected by PM_SLEEP
kernel/power/Kconfig:104: symbol PM_SLEEP depends on HIBERNATE_CALLBACKS
kernel/power/Kconfig:31: symbol HIBERNATE_CALLBACKS is selected by
HIBERNATION
kernel/power/Kconfig:34: symbol HIBERNATION depends on SWAP
init/Kconfig:250: symbol SWAP depends on BLOCK
block/Kconfig:5: symbol BLOCK is selected by UBIFS_FS
fs/ubifs/Kconfig:1: symbol UBIFS_FS depends on MISC_FILESYSTEMS
fs/Kconfig:220: symbol MISC_FILESYSTEMS is selected by ACPI_APEI
drivers/acpi/apei/Kconfig:8: symbol ACPI_APEI depends on ACPI
drivers/acpi/Kconfig:9: symbol ACPI depends on ARCH_SUPPORTS_ACPI
<<<< LOOK HERE
drivers/acpi/Kconfig:6: symbol ARCH_SUPPORTS_ACPI is selected by
IA64_HP_SIM
arch/ia64/Kconfig:200: symbol IA64_HP_SIM is part of choice <choice>

At any rate, a 3 mn git bisect tells me the circular dependency is exposed by this change:

f3fd6cd74fedf99b6060f75df00943fda13b65f2 is the first bad commit commit f3fd6cd74fedf99b6060f75df00943fda13b65f2

Author: Chandan Rajendra <[email protected]>
Date:   Sat Dec 8 12:21:38 2018 +0530

fscrypt: remove filesystem specific build config option

In order to have a common code base for fscrypt "post read" processing for all filesystems which support encryption, this commit removes filesystem specific build config option (e.g.
CONFIG_EXT4_FS_ENCRYPTION)
and replaces it with a build option (i.e. CONFIG_FS_ENCRYPTION) whose value affects all the filesystems making use of fscrypt.

Quoted 3 lines

FWIW, The patch at https://patchwork.kernel.org/patch/10725883/ fixes this problem by removing "select BLOCK if FS_ENCRYPTION" from fs/ubifs/Kconfig.

Quoted 9 lines
Quoted 4 lines

OK

Pierre-Louis, can you check if this patch makes the issue go away for you, please?

Wondering if Chandan provided the right pointer, I wasn't able to apply this patch, but commenting out "select BLOCK if FS_ENCRYPTION" in fs/ubifs/Kconfig makes the circular dependency go away. All good for me.

V4 version of the patchset (which is part of linux-next) had introduced the "select BLOCK if FS_ENCRYPTION". The V5 patchset omits that line. Hence the patch did not apply cleanly. However just commenting out that line was the right thing to do.

Signature · 1 line
chandan
maintainer
#30 lore raw

Re: [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

In reply to Sinan Kaya
Quoted 6 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This code relies on IOSF_MBI and IOSF_MBI depends on PCI. For this reason, add a direct dependency on CONFIG_PCI to the IOSF_MBI driver.

I still don't understand what's going on with dependencies here and still don't have the cover letter or anything :( . As far as I can tell the above commit is in Linus' tree so I'd expect I can just apply it directly but you were saying that this needs to go via some other tree so I'm a bit confused as to what's going on here.

#31 lore raw

Re: [PATCH v5 08/11] ASoC: Intel: atom: Make PCI dependency explicit

Quoted 10 lines · Mark Brown
Quoted 6 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This code relies on IOSF_MBI and IOSF_MBI depends on PCI. For this reason, add a direct dependency on CONFIG_PCI to the IOSF_MBI driver.

Quoted 4 lines

I still don't understand what's going on with dependencies here and still don't have the cover letter or anything :( . As far as I can tell the above commit is in Linus' tree so I'd expect I can just apply it directly

Yes, you can.

Quoted 2 lines

but you were saying that this needs to go via some other tree so I'm a bit confused as to what's going on here.

I guess Sinan wanted it to go in via the ACPI tree like 5d32a66541c4, but that of course isn't necessary.

If you want to pick up anything from this series, please do. I will pick up what's left. :-)

author
#32 lore raw

PATCH v5 9/11 mmc: sdhci-acpi: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver relies on IOSF_MBI and IOSF_MBI depends on PCI. For this reason, add a direct dependency to CONFIG_PCI here.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/mmc/host/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mmc/host/Kconfig +1 -1
index e26b8145efb3..1b9401fe94c0 100644
@@ -116,7 +116,7 @@ config MMC_RICOH_MMC
 
 config MMC_SDHCI_ACPI
 	tristate "SDHCI support for ACPI enumerated SDHCI controllers"
-	depends on MMC_SDHCI && ACPI
+	depends on MMC_SDHCI && ACPI && PCI
 	select IOSF_MBI if X86
 	help
 	  This selects support for ACPI enumerated SDHCI controllers,
Signature · 1 line
2.19.0
#33 lore raw

Re: [PATCH v5 09/11] mmc: sdhci-acpi: Make PCI dependency explicit

Quoted 5 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver relies on IOSF_MBI and IOSF_MBI depends on PCI. For this reason, add a direct dependency to CONFIG_PCI here.

This driver does not rely on IOSF_MBI

You wrote on the other email QCOM doesn't use PCI but here you are making them depend on it. Please explain.

Quoted 21 lines
Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/mmc/host/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mmc/host/Kconfig +1 -1
index e26b8145efb3..1b9401fe94c0 100644
@@ -116,7 +116,7 @@ config MMC_RICOH_MMC
 
 config MMC_SDHCI_ACPI
 	tristate "SDHCI support for ACPI enumerated SDHCI controllers"
-	depends on MMC_SDHCI && ACPI
+	depends on MMC_SDHCI && ACPI && PCI
 	select IOSF_MBI if X86
 	help
 	  This selects support for ACPI enumerated SDHCI controllers,
author
#34 lore raw

Re: [PATCH v5 09/11] mmc: sdhci-acpi: Make PCI dependency explicit

Quoted 10 lines · Adrian Hunter
Quoted 5 lines · Sinan Kaya

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. This driver relies on IOSF_MBI and IOSF_MBI depends on PCI. For this reason, add a direct dependency to CONFIG_PCI here.

This driver does not rely on IOSF_MBI

X86 version does.

select IOSF_MBI if X86

Quoted 3 lines

You wrote on the other email QCOM doesn't use PCI but here you are making them depend on it. Please explain.

I matched the behavior before my change. This code wouldn't be available without PCI and ACPI on previous kernel versions and left kconfig cleanup as another exercise for some other day.

Quoted 22 lines
Quoted 21 lines
Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/mmc/host/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mmc/host/Kconfig +1 -1
index e26b8145efb3..1b9401fe94c0 100644
@@ -116,7 +116,7 @@ config MMC_RICOH_MMC
 
 config MMC_SDHCI_ACPI
      tristate "SDHCI support for ACPI enumerated SDHCI controllers"
-     depends on MMC_SDHCI && ACPI
+     depends on MMC_SDHCI && ACPI && PCI
      select IOSF_MBI if X86
      help
        This selects support for ACPI enumerated SDHCI controllers,
author
#35 lore raw

PATCH v5 10/11 x86/intel/lpss: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. LPSS code relies on PCI infrastructure but this dependency has not been explicitly called out.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 arch/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/Kconfig +1 -1
index cc4f0c6ee1ed..322a58ababb0 100644
@@ -617,7 +617,7 @@ config X86_INTEL_QUARK
 
 config X86_INTEL_LPSS
 	bool "Intel Low Power Subsystem Support"
-	depends on X86 && ACPI
+	depends on X86 && ACPI && PCI
 	select COMMON_CLK
 	select PINCTRL
 	select IOSF_MBI
Signature · 1 line
2.19.0
#36 lore raw

[tip:x86/urgent] x86/intel/lpss: Make PCI dependency explicit

x86/intel/lpss: Make PCI dependency explicit
Commit-ID:  5962dd22f0ff6f7d72fff974b3c637d52586643e
Gitweb:     https://git.kernel.org/tip/5962dd22f0ff6f7d72fff974b3c637d52586643e
Author:     Sinan Kaya <[email protected]>
AuthorDate: Wed, 2 Jan 2019 18:10:37 +0000
Committer:  Borislav Petkov <[email protected]>
CommitDate: Fri, 11 Jan 2019 19:32:22 +0100

x86/intel/lpss: Make PCI dependency explicit

After commit

5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")

dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. LPSS code relies on PCI infrastructure but this dependency has not been explicitly called out so do that.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
Cc "H. Peter Anvin" <[email protected]>
Cc x86-ml <[email protected]>
---
 arch/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/Kconfig +1 -1
index 15af091611e2..4b4a7f32b68e 100644
@@ -617,7 +617,7 @@ config X86_INTEL_QUARK
 
 config X86_INTEL_LPSS
 	bool "Intel Low Power Subsystem Support"
-	depends on X86 && ACPI
+	depends on X86 && ACPI && PCI
 	select COMMON_CLK
 	select PINCTRL
 	select IOSF_MBI
author
#37 lore raw

PATCH v5 11/11 drivers: thermal: int340x_thermal: Make PCI dependency explicit

In reply to a message not in the archive

After 'commit 5d32a66541c4 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")' dependencies on CONFIG_PCI that previously were satisfied implicitly through dependencies on CONFIG_ACPI have to be specified directly. IOSF_CORE depends on PCI. For this reason, add a direct dependency on CONFIG_PCI.

Fixes 5d32a66541c46 ("PCI/ACPI: Allow ACPI to be built without CONFIG_PCI set")
---
 drivers/thermal/intel/int340x_thermal/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/thermal/intel/int340x_thermal/Kconfig +1 -1
index fba1976d5f8d..2acddf8fb314 100644
@@ -4,7 +4,7 @@
 
 config INT340X_THERMAL
 	tristate "ACPI INT340X thermal drivers"
-	depends on X86 && ACPI
+	depends on X86 && ACPI && PCI
 	select THERMAL_GOV_USER_SPACE
 	select ACPI_THERMAL_REL
 	select ACPI_FAN
Signature · 1 line
2.19.0