Skip to contentSkip to conversation

platform: ioremap: Build iomap functions even without HAS_IOMEM

RFC PATCH In review David Gow 4 participants 4 messages Jul 21, 2020 – Jul 23, 2020

Subsystems (3)
lore
Share & tools
lore

David Gow

author
#0 lore raw

RFC PATCH 1/1 platform: ioremap: Build iomap functions even without HAS_IOMEM

A number of drivers use devm_platform_ioremap_resource(), but do not (or did not) explicitly depend on IOMEM[1,2]. Given that the only platform without HAS_IOMEM seems to be UML, and it has sufficient stubs for devm_platform_ioremap_resource() and its dependencies to build, we can remove the HAS_IOMEM requirement here, rather than playing whack-a-mole with different drivers which don't try to build against ARCH=um.

The reason this works at the moment is that stub ioremap and iounmap functions were added to UML to support this sort-of thing[3]. This particular change doesn't require adding any additional stubs, but there is possibly room in the future to stub out the remaining iomem functions (or to provide mock implementations for testing), and get rid of HAS_IOMEM entirely.

[1]: https://lkml.org/lkml/2020/6/30/176
[2]:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1e7468bd9d30a21e059af477106dc5588ae52dff
[3]:
https://www.mail-archive.com/[email protected]/msg1405627.html

---
 drivers/base/platform.c | 2 --
 lib/Makefile            | 2 +-
 2 files changed, 1 insertion(+), 3 deletions(-)
drivers/base/platform.c +0 -2
index c0d0a5490ac6..628dde6675cf 100644
@@ -61,7 +61,6 @@ struct resource *platform_get_resource(struct platform_device *dev,
 }
 EXPORT_SYMBOL_GPL(platform_get_resource);
 
-#ifdef CONFIG_HAS_IOMEM
 /**
  * devm_platform_get_and_ioremap_resource - call devm_ioremap_resource() for a
  *					    platform device and get resource
@@ -135,7 +134,6 @@ devm_platform_ioremap_resource_byname(struct platform_device *pdev,
 	return devm_ioremap_resource(&pdev->dev, res);
 }
 EXPORT_SYMBOL_GPL(devm_platform_ioremap_resource_byname);
-#endif /* CONFIG_HAS_IOMEM */
 
 /**
  * platform_get_irq_optional - get an optional IRQ for a device
lib/Makefile +1 -1
index b1c42c10073b..35c21af33b93 100644
@@ -115,7 +115,7 @@ obj-y += math/ crypto/
 
 obj-$(CONFIG_GENERIC_IOMAP) += iomap.o
 obj-$(CONFIG_GENERIC_PCI_IOMAP) += pci_iomap.o
-obj-$(CONFIG_HAS_IOMEM) += iomap_copy.o devres.o
+obj-y += iomap_copy.o devres.o
 obj-$(CONFIG_CHECK_SIGNATURE) += check_signature.o
 obj-$(CONFIG_DEBUG_LOCKING_API_SELFTESTS) += locking-selftest.o
 
Signature · 1 line
2.28.0.rc0.105.gf9edc3c819-goog
Reviewed-by on v1 1/1
#1 lore raw
Quoted 15 lines · David Gow

A number of drivers use devm_platform_ioremap_resource(), but do not (or did not) explicitly depend on IOMEM[1,2]. Given that the only platform without HAS_IOMEM seems to be UML, and it has sufficient stubs for devm_platform_ioremap_resource() and its dependencies to build, we can remove the HAS_IOMEM requirement here, rather than playing whack-a-mole with different drivers which don't try to build against ARCH=um.

The reason this works at the moment is that stub ioremap and iounmap functions were added to UML to support this sort-of thing[3]. This particular change doesn't require adding any additional stubs, but there is possibly room in the future to stub out the remaining iomem functions (or to provide mock implementations for testing), and get rid of HAS_IOMEM entirely.

Quoted 7 lines

[1]: https://lkml.org/lkml/2020/6/30/176
[2]:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1e7468bd9d30a21e059af477106dc5588ae52dff
[3]:
https://www.mail-archive.com/[email protected]/msg1405627.html

#2 lore raw
In reply to David Gow
Quoted 14 lines · On Mon, 20 Jul 2020 22:45:28 -0700 David Gow <[email protected]> wrote:

A number of drivers use devm_platform_ioremap_resource(), but do not (or did not) explicitly depend on IOMEM[1,2]. Given that the only platform without HAS_IOMEM seems to be UML, and it has sufficient stubs for devm_platform_ioremap_resource() and its dependencies to build, we can remove the HAS_IOMEM requirement here, rather than playing whack-a-mole with different drivers which don't try to build against ARCH=um.

The reason this works at the moment is that stub ioremap and iounmap functions were added to UML to support this sort-of thing[3]. This particular change doesn't require adding any additional stubs, but there is possibly room in the future to stub out the remaining iomem functions (or to provide mock implementations for testing), and get rid of HAS_IOMEM entirely.

Quoted 5 lines

[1]: https://lkml.org/lkml/2020/6/30/176
[2]:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1e7468bd9d30a21e059af477106dc5588ae52dff
[3]:
https://www.mail-archive.com/[email protected]/msg1405627.html

Hi David,

I don't quite follow why we change when iomap_copy.c is built. Was this just a case of there not seeming to be any reason to protect it or is there a direct dependency on something in there that I'm missing?

Otherwise looks good to me.

Jonathan

Quoted 40 lines
---
 drivers/base/platform.c | 2 --
 lib/Makefile            | 2 +-
 2 files changed, 1 insertion(+), 3 deletions(-)
drivers/base/platform.c +0 -2
index c0d0a5490ac6..628dde6675cf 100644
@@ -61,7 +61,6 @@ struct resource *platform_get_resource(struct platform_device *dev,
 }
 EXPORT_SYMBOL_GPL(platform_get_resource);
 
-#ifdef CONFIG_HAS_IOMEM
 /**
  * devm_platform_get_and_ioremap_resource - call devm_ioremap_resource() for a
  *					    platform device and get resource
@@ -135,7 +134,6 @@ devm_platform_ioremap_resource_byname(struct platform_device *pdev,
 	return devm_ioremap_resource(&pdev->dev, res);
 }
 EXPORT_SYMBOL_GPL(devm_platform_ioremap_resource_byname);
-#endif /* CONFIG_HAS_IOMEM */
 
 /**
  * platform_get_irq_optional - get an optional IRQ for a device
lib/Makefile +1 -1
index b1c42c10073b..35c21af33b93 100644
@@ -115,7 +115,7 @@ obj-y += math/ crypto/
 
 obj-$(CONFIG_GENERIC_IOMAP) += iomap.o
 obj-$(CONFIG_GENERIC_PCI_IOMAP) += pci_iomap.o
-obj-$(CONFIG_HAS_IOMEM) += iomap_copy.o devres.o
+obj-y += iomap_copy.o devres.o
 obj-$(CONFIG_CHECK_SIGNATURE) += check_signature.o
 obj-$(CONFIG_DEBUG_LOCKING_API_SELFTESTS) += locking-selftest.o
 
#3 lore raw
In reply to David Gow
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in https://git-scm.com/docs/git-format-patch]

Hi David,

[FYI, it's a private test report for your RFC patch.]
[auto build test ERROR on driver-core/driver-core-testing] [also build test ERROR on linux/master linus/master v5.8-rc6 next-20200722] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/David-Gow/platform-ioremap-Build-iomap-functions-even-without-HAS_IOMEM/20200721-134709
base:   https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core.git 6bdb486c5a628f7a927c2658166e3a5ef1f883e7
config: s390-allnoconfig (attached as .config)
compiler: s390-linux-gcc (GCC) 9.3.0

reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=s390

If you fix the issue, kindly add following tag as appropriate

All errors (new ones prefixed by >>):

s390-linux-ld: lib/devres.o: in function `devm_ioremap_release':

>> devres.c:(.text+0x28): undefined reference to `iounmap'
   s390-linux-ld: lib/devres.o: in function `__devm_ioremap':
>> devres.c:(.text+0xa2): undefined reference to `ioremap'

s390-linux-ld: lib/devres.o: in function `devm_iounmap':
devres.c:(.text+0x284): undefined reference to `iounmap'

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org