Skip to contentSkip to conversation

kunit,um: Fix kunit.py build --alltests

PATCH In review David Gow 5 participants 12 messages Feb 18, 2022 – Feb 19, 2022

Subsystems (9)
lore
Share & tools
lore

David Gow

author
#0 lore raw

cover kunit,um: Fix kunit.py build --alltests

kunit_tool's --alltests option uses UML and make allyesconfig to produce a configuration which enables as many tests as possible. However, make ARCH=um allyesconfig is broken for a number of reasons.

Fix a few different UML build breakages, and disable a few config options in kunit_tool in order to get this kernel compiling again.

Note that the resulting kernel still doesn't run, but having it compile is the first step to fixing that.

David Gow (3):
drm/amdgpu: Make smu7_hwmgr build on UML IB/qib: Compile under User-Mode Linux
kunit: tool: Disable broken options for --alltests

Randy Dunlap (1):
drm/amdgpu: Fix compilation under UML

 drivers/gpu/drm/amd/amdkfd/kfd_crat.c               | 6 +++---
 drivers/gpu/drm/amd/amdkfd/kfd_topology.c           | 2 +-
 drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c | 2 +-
 drivers/infiniband/hw/qib/qib_wc_x86_64.c           | 4 ++++
 tools/testing/kunit/configs/broken_on_uml.config    | 5 +++++
 5 files changed, 14 insertions(+), 5 deletions(-)
Signature · 1 line
2.35.1.265.g69c8d7142f-goog
author
#1 lore raw

PATCH 1/4 drm/amdgpu: Fix compilation under UML

From: Randy Dunlap <[email protected]>

cpuinfo_x86 and its associated macros are not available under ARCH=um, even though CONFIG_X86_64 is defined.

This patch (and discussion) were originally posted here:
https://lkml.org/lkml/2022/1/24/1547

This produces the following build errors:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1556:9: note: in expansion of macro ‘cpu_data’

  return cpu_data(first_cpu_of_numa_node).apicid;
         ^~~~~~~~

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1560:1: error: control reaches end of non-void function [-Werror=return-type]

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_fill_iolink_info_for_cpu’:
../arch/um/include/asm/processor-generic.h:103:19: error: called object is not a function or function pointer

 #define cpu_data (&boot_cpu_data)
                  ~^~~~~~~~~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1688:27: note: in expansion of macro ‘cpu_data’
  struct cpuinfo_x86 *c = &cpu_data(0);
                           ^~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:7: error: dereferencing pointer to incomplete type ‘struct cpuinfo_x86’
  if (c->x86_vendor == X86_VENDOR_AMD)
       ^~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:23: error: ‘X86_VENDOR_AMD’ undeclared (first use in this function); did you mean ‘X86_VENDOR_ANY’?
  if (c->x86_vendor == X86_VENDOR_AMD)
                       ^~~~~~~~~~~~~~
                       X86_VENDOR_ANY

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_create_vcrat_image_cpu’:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1742:11: warning: unused variable ‘entries’ [-Wunused-variable]
uint32_t entries = 0;

---
 drivers/gpu/drm/amd/amdkfd/kfd_crat.c     | 6 +++---
 drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +3 -3
index 9624bbe8b501..b1e2d117be3d 100644
@@ -1682,7 +1682,7 @@ static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size,
 	return 0;
 }
 
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
 static int kfd_fill_iolink_info_for_cpu(int numa_node_id, int *avail_size,
 				uint32_t *num_entries,
 				struct crat_subtype_iolink *sub_type_hdr)
@@ -1741,7 +1741,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
 	struct crat_subtype_generic *sub_type_hdr;
 	int avail_size = *size;
 	int numa_node_id;
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
 	uint32_t entries = 0;
 #endif
 	int ret = 0;
@@ -1806,7 +1806,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
 			sub_type_hdr->length);
 
 		/* Fill in Subtype: IO Link */
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
 		ret = kfd_fill_iolink_info_for_cpu(numa_node_id, &avail_size,
 				&entries,
 				(struct crat_subtype_iolink *)sub_type_hdr);
drivers/gpu/drm/amd/amdkfd/kfd_topology.c +1 -1
index 948fbb39336e..b38fc530ffe2 100644
@@ -1552,7 +1552,7 @@ static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask)
 	first_cpu_of_numa_node = cpumask_first(cpumask);
 	if (first_cpu_of_numa_node >= nr_cpu_ids)
 		return -1;
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
 	return cpu_data(first_cpu_of_numa_node).apicid;
 #else
 	return first_cpu_of_numa_node;
Signature · 7 lines
2.35.1.265.g69c8d7142f-goog


_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
maintainer
#2 lore raw

Re: [PATCH 1/4] drm/amdgpu: Fix compilation under UML

Am 2022-02-18 um 02:57 schrieb David Gow:

Quoted 45 lines

From: Randy Dunlap <[email protected]>

cpuinfo_x86 and its associated macros are not available under ARCH=um, even though CONFIG_X86_64 is defined.

This patch (and discussion) were originally posted here:
https://lkml.org/lkml/2022/1/24/1547

This produces the following build errors:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1556:9: note: in expansion of macro ‘cpu_data’

   return cpu_data(first_cpu_of_numa_node).apicid;
          ^~~~~~~~

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1560:1: error: control reaches end of non-void function [-Werror=return-type]

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_fill_iolink_info_for_cpu’:
../arch/um/include/asm/processor-generic.h:103:19: error: called object is not a function or function pointer

  #define cpu_data (&boot_cpu_data)
                   ~^~~~~~~~~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1688:27: note: in expansion of macro ‘cpu_data’
   struct cpuinfo_x86 *c = &cpu_data(0);
                            ^~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:7: error: dereferencing pointer to incomplete type ‘struct cpuinfo_x86’
   if (c->x86_vendor == X86_VENDOR_AMD)
        ^~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:23: error: ‘X86_VENDOR_AMD’ undeclared (first use in this function); did you mean ‘X86_VENDOR_ANY’?
   if (c->x86_vendor == X86_VENDOR_AMD)
                        ^~~~~~~~~~~~~~
                        X86_VENDOR_ANY

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_create_vcrat_image_cpu’:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1742:11: warning: unused variable ‘entries’ [-Wunused-variable]
uint32_t entries = 0;

---
  drivers/gpu/drm/amd/amdkfd/kfd_crat.c     | 6 +++---
  drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 2 +-
  2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +0 -0
index 9624bbe8b501..b1e2d117be3d 100644

drivers/gpu/drm/amd/amdkfd/kfd_crat.c static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size, · line 1685

Quoted 5 lines
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +1 -1
@@ -1682,7 +1682,7 @@ static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size,
  	return 0;
  }
 
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)

I don't think it makes sense to compile a hardware device driver in a UML config. Instead of scattering UML #ifdefs through our code, I would recommend adding this to Kconfig:

drivers/gpu/drm/Kconfig +1 -1
@@ -254,7 +254,7 @@ source "drivers/gpu/drm/radeon/Kconfig"
 
 config DRM_AMDGPU
        tristate "AMD GPU"
-       depends on DRM && PCI && MMU
+       depends on DRM && PCI && MMU && !UML
        select FW_LOADER
        select DRM_KMS_HELPER
        select DRM_SCHED

That would address patch 2 of this series as well.

Regards,
  Felix

Quoted 34 lines
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +2 -2
  static int kfd_fill_iolink_info_for_cpu(int numa_node_id, int *avail_size,
  				uint32_t *num_entries,
  				struct crat_subtype_iolink *sub_type_hdr)
@@ -1741,7 +1741,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
  	struct crat_subtype_generic *sub_type_hdr;
  	int avail_size = *size;
  	int numa_node_id;
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
  	uint32_t entries = 0;
  #endif
  	int ret = 0;
@@ -1806,7 +1806,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
  			sub_type_hdr->length);
 
  		/* Fill in Subtype: IO Link */
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
  		ret = kfd_fill_iolink_info_for_cpu(numa_node_id, &avail_size,
  				&entries,
  				(struct crat_subtype_iolink *)sub_type_hdr);
drivers/gpu/drm/amd/amdkfd/kfd_topology.c +1 -1
index 948fbb39336e..b38fc530ffe2 100644
@@ -1552,7 +1552,7 @@ static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask)
  	first_cpu_of_numa_node = cpumask_first(cpumask);
  	if (first_cpu_of_numa_node >= nr_cpu_ids)
  		return -1;
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
  	return cpu_data(first_cpu_of_numa_node).apicid;
  #else
  	return first_cpu_of_numa_node;
#3 lore raw

Re: [PATCH 1/4] drm/amdgpu: Fix compilation under UML

Quoted 70 lines · Felix Kuehling

Am 2022-02-18 um 02:57 schrieb David Gow:

Quoted 45 lines

From: Randy Dunlap <[email protected]>

cpuinfo_x86 and its associated macros are not available under ARCH=um, even though CONFIG_X86_64 is defined.

This patch (and discussion) were originally posted here:
https://lkml.org/lkml/2022/1/24/1547

This produces the following build errors:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1556:9: note: in expansion of macro ‘cpu_data’

   return cpu_data(first_cpu_of_numa_node).apicid;
          ^~~~~~~~

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1560:1: error: control reaches end of non-void function [-Werror=return-type]

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_fill_iolink_info_for_cpu’:
../arch/um/include/asm/processor-generic.h:103:19: error: called object is not a function or function pointer

  #define cpu_data (&boot_cpu_data)
                   ~^~~~~~~~~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1688:27: note: in expansion of macro ‘cpu_data’
   struct cpuinfo_x86 *c = &cpu_data(0);
                            ^~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:7: error: dereferencing pointer to incomplete type ‘struct cpuinfo_x86’
   if (c->x86_vendor == X86_VENDOR_AMD)
        ^~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:23: error: ‘X86_VENDOR_AMD’ undeclared (first use in this function); did you mean ‘X86_VENDOR_ANY’?
   if (c->x86_vendor == X86_VENDOR_AMD)
                        ^~~~~~~~~~~~~~
                        X86_VENDOR_ANY

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_create_vcrat_image_cpu’:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1742:11: warning: unused variable ‘entries’ [-Wunused-variable]
uint32_t entries = 0;

---
  drivers/gpu/drm/amd/amdkfd/kfd_crat.c     | 6 +++---
  drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 2 +-
  2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +0 -0
index 9624bbe8b501..b1e2d117be3d 100644

drivers/gpu/drm/amd/amdkfd/kfd_crat.c static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size, · line 1685

Quoted 5 lines
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +1 -1
@@ -1682,7 +1682,7 @@ static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size,
      return 0;
  }
 
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)

I don't think it makes sense to compile a hardware device driver in a UML config. Instead of scattering UML #ifdefs through our code, I would recommend adding this to Kconfig:

drivers/gpu/drm/Kconfig +1 -1
@@ -254,7 +254,7 @@ source "drivers/gpu/drm/radeon/Kconfig"
 
  config DRM_AMDGPU
         tristate "AMD GPU"
-       depends on DRM && PCI && MMU
+       depends on DRM && PCI && MMU && !UML
         select FW_LOADER
         select DRM_KMS_HELPER
         select DRM_SCHED
Quoted 1 line

That would address patch 2 of this series as well.

I agree. Otherwise, we are always going to be chasing these issues.

Alex

Quoted 39 lines

Regards,
Felix

Quoted 34 lines
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +2 -2
  static int kfd_fill_iolink_info_for_cpu(int numa_node_id, int *avail_size,
                              uint32_t *num_entries,
                              struct crat_subtype_iolink *sub_type_hdr)
@@ -1741,7 +1741,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
      struct crat_subtype_generic *sub_type_hdr;
      int avail_size = *size;
      int numa_node_id;
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
      uint32_t entries = 0;
  #endif
      int ret = 0;
@@ -1806,7 +1806,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
                      sub_type_hdr->length);
 
              /* Fill in Subtype: IO Link */
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
              ret = kfd_fill_iolink_info_for_cpu(numa_node_id, &avail_size,
                              &entries,
                              (struct crat_subtype_iolink *)sub_type_hdr);
drivers/gpu/drm/amd/amdkfd/kfd_topology.c +1 -1
index 948fbb39336e..b38fc530ffe2 100644
@@ -1552,7 +1552,7 @@ static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask)
      first_cpu_of_numa_node = cpumask_first(cpumask);
      if (first_cpu_of_numa_node >= nr_cpu_ids)
              return -1;
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)
      return cpu_data(first_cpu_of_numa_node).apicid;
  #else
      return first_cpu_of_numa_node;
author
#4 lore raw

Re: [PATCH 1/4] drm/amdgpu: Fix compilation under UML

In reply to Felix Kuehling
Quoted 51 lines · Felix Kuehling

Am 2022-02-18 um 02:57 schrieb David Gow:

Quoted 48 lines

From: Randy Dunlap <[email protected]>

cpuinfo_x86 and its associated macros are not available under ARCH=um, even though CONFIG_X86_64 is defined.

This patch (and discussion) were originally posted here:
https://lkml.org/lkml/2022/1/24/1547

This produces the following build errors:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1556:9: note: in expansion of macro ‘cpu_data’

   return cpu_data(first_cpu_of_numa_node).apicid;
          ^~~~~~~~

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_topology.c:1560:1: error: control reaches end of non-void function [-Werror=return-type]

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_fill_iolink_info_for_cpu’:
../arch/um/include/asm/processor-generic.h:103:19: error: called object is not a function or function pointer

  #define cpu_data (&boot_cpu_data)
                   ~^~~~~~~~~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1688:27: note: in expansion of macro ‘cpu_data’
   struct cpuinfo_x86 *c = &cpu_data(0);
                            ^~~~~~~~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:7: error: dereferencing pointer to incomplete type ‘struct cpuinfo_x86’
   if (c->x86_vendor == X86_VENDOR_AMD)
        ^~
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1691:23: error: ‘X86_VENDOR_AMD’ undeclared (first use in this function); did you mean ‘X86_VENDOR_ANY’?
   if (c->x86_vendor == X86_VENDOR_AMD)
                        ^~~~~~~~~~~~~~
                        X86_VENDOR_ANY

../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c: In function ‘kfd_create_vcrat_image_cpu’:
../drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_crat.c:1742:11: warning: unused variable ‘entries’ [-Wunused-variable]
uint32_t entries = 0;

---
  drivers/gpu/drm/amd/amdkfd/kfd_crat.c     | 6 +++---
  drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 2 +-
  2 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +0 -0
index 9624bbe8b501..b1e2d117be3d 100644
@@ -1682,7 +1682,7 @@ static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size,
      return 0;
  }
 
Quoted 7 lines

drivers/gpu/drm/amd/amdkfd/kfd_crat.c static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size, · line 1685

Quoted 2 lines
drivers/gpu/drm/amd/amdkfd/kfd_crat.c +1 -1
-#ifdef CONFIG_X86_64
+#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)

I don't think it makes sense to compile a hardware device driver in a UML config. Instead of scattering UML #ifdefs through our code, I would recommend adding this to Kconfig:

There are cases where I think it could make sense to have a hardware driver under UML, though I agree they're pretty rare.

In particular, there's the virtio PCI support in UML, which one could potentially hook up a GPU to. The case I care more about is the ability to run KUnit tests under UML: if amdgpu wanted to have KUnit tests, it could still run them in qemu (or on real hardware), but UML is faster and more convenient, if the code being tested can compile under it.

So I have a slight preference personally for fixing this, to unblock those uses.

That being said, it's definitely not worth placing a significant burden on you maintaining these things if no-one uses them. And since there doesn't appear to be any such use at the moment[1], I've no strong objection to just disabling this for now (it can always be re-enabled and fixed if it becomes useful later).

Cheers,
-- David

[1] The proposed DRM KUnit tests don't require any actual hardware drivers, as I understand it:
https://lore.kernel.org/all/[email protected]/T/

author
#5 lore raw

PATCH 2/4 drm/amdgpu: Make smu7_hwmgr build on UML

In reply to David Gow

The User-Mode-Linux architecture (with the x86_64 subarch) defines CONFIG_X86_64, but doesn't expose the cpuinfo_x86 struct (instead there's a cpuinfo_um struct).

In order to allow UML to build with allyesconfig, only check cpuinfo_x86 on non-UML architectures.

Fixes b3dc549986 ("mdgpu: Disable PCIE_DPM on Intel RKL Platform")
---
 drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c +1 -1
index a1e11037831a..a162552f7845 100644
@@ -1738,7 +1738,7 @@ static int smu7_disable_dpm_tasks(struct pp_hwmgr *hwmgr)
 
 static bool intel_core_rkl_chk(void)
 {
-#if IS_ENABLED(CONFIG_X86_64)
+#if IS_ENABLED(CONFIG_X86_64) && !defined(CONFIG_UML)
 	struct cpuinfo_x86 *c = &cpu_data(0);
 
 	return (c->x86 == 6 && c->x86_model == INTEL_FAM6_ROCKETLAKE);
Signature · 1 line
2.35.1.265.g69c8d7142f-goog
author
#6 lore raw

PATCH 3/4 IB/qib: Compile under User-Mode Linux

In reply to David Gow

The User-Mode-Linux architecture (with the x86_64 subarch) defines CONFIG_X86_64, but doesn't expose the cpuinfo_x86 struct (instead there's a cpuinfo_um struct).

In order to allow UML to build with allyesconfig, only check cpuinfo_x86 on non-UML architectures.

---
 drivers/infiniband/hw/qib/qib_wc_x86_64.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/infiniband/hw/qib/qib_wc_x86_64.c +4 -0
index edd0ddbd4481..76fef1321c26 100644
@@ -146,5 +146,9 @@ void qib_disable_wc(struct qib_devdata *dd)
  */
 int qib_unordered_wc(void)
 {
+#ifndef CONFIG_UML
 	return boot_cpu_data.x86_vendor != X86_VENDOR_AMD;
+#else
+	return 0;
+#endif
 }
Signature · 1 line
2.35.1.265.g69c8d7142f-goog
author
#7 lore raw

PATCH 4/4 kunit: tool: Disable broken options for --alltests

In reply to David Gow

There are a number of Kconfig options which break compilation under UML with allyesconfig. As kunit_tool's --alltests option is based on allyesconfig and UML, we need to update the list of broken options to make --alltests build again.

Note that, while this does build again, it still segfaults on startup, so more work remains to be done.

They are:
- CONFIG_VFIO_PCI: Needs ioport_map/ioport_unmap.
- CONFIG_INFINIBAND_RDMAVT: Needs cpuinfo_x86 and __copy_user_nocache
- CONFIG_BNXT: Failing under UML with -Werror
ERROR:root:../drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c: In function ‘bnxt_ptp_enable’:
../drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c:400:43: error: array subscript 255 is above array bounds of ‘struct pps_pin[4]’ [-Werror=array-bounds]

  400 |                         ptp->pps_info.pins[pin_id].event = BNXT_PPS_EVENT_EXTERNAL;
      |                         ~~~~~~~~~~~~~~~~~~^~~~~~~~

- CONFIG_PATA_CS5535: Needs MSR access (__tracepoint_{read,write}_msr)
- CONFIG_VDPA: Enables CONFIG_DMA_OPS, which is unimplemented. ('dma_ops' is not defined)

These are all issues which should be investigated properly and the corresponding options either fixed or disabled under UML. Having this list of broken options should act as a good to-do list here, and will allow these issues to be worked on independently, and other tests to work in the meantime.

---
 tools/testing/kunit/configs/broken_on_uml.config | 5 +++++
 1 file changed, 5 insertions(+)
tools/testing/kunit/configs/broken_on_uml.config +5 -0
index 690870043ac0..546482b0bc4d 100644
@@ -42,3 +42,8 @@
 # CONFIG_ADI_AXI_ADC is not set
 # CONFIG_DEBUG_PAGEALLOC is not set
 # CONFIG_PAGE_POISONING is not set
+# CONFIG_VFIO_PCI is not set
+# CONFIG_INFINIBAND_RDMAVT is not set
+# CONFIG_BNXT is not set
+# CONFIG_PATA_CS5535 is not set
+# CONFIG_VDPA is not set
Signature · 1 line
2.35.1.265.g69c8d7142f-goog
maintainer
#8 lore raw

Re: [PATCH 4/4] kunit: tool: Disable broken options for --alltests

Quoted 3 lines · David Gow

Note that, while this does build again, it still segfaults on startup, so more work remains to be done.

That's probably just a lot more stuff getting included somehow?

Quoted 8 lines

They are:
- CONFIG_VFIO_PCI: Needs ioport_map/ioport_unmap.
- CONFIG_INFINIBAND_RDMAVT: Needs cpuinfo_x86 and __copy_user_nocache
- CONFIG_BNXT: Failing under UML with -Werror
ERROR:root:../drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c: In function ‘bnxt_ptp_enable’:
../drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c:400:43: error: array subscript 255 is above array bounds of ‘struct pps_pin[4]’ [-Werror=array-bounds]

  400 |                         ptp->pps_info.pins[pin_id].event = BNXT_PPS_EVENT_EXTERNAL;
      |                         ~~~~~~~~~~~~~~~~~~^~~~~~~~
Quoted 9 lines

- CONFIG_PATA_CS5535: Needs MSR access (__tracepoint_{read,write}_msr)
- CONFIG_VDPA: Enables CONFIG_DMA_OPS, which is unimplemented. ('dma_ops' is not defined)

These are all issues which should be investigated properly and the corresponding options either fixed or disabled under UML. Having this list of broken options should act as a good to-do list here, and will allow these issues to be worked on independently, and other tests to work in the meantime.

I'm not really sure it makes sense to even do anything other than disabling these.

It looks like all of them are just exposed by now being able to build PCI drivers on UML. Surely the people writing the driver didn't expect their drivers to run over simulated PCI (which is what the UML PCI support is all about).

Now from a PCI driver point of view you can't really tell the difference (and anyway the driver won't be probed), but the issues (at least the build time ones) come from having

    UML && PCI && X86_64

or

    UML && PCI && X86_32

because drivers typically depend on X86_64 or X86_32, rather than on "X86 && X86_64" or "X86 && X86_32". In a sense thus, the issue is those drivers don't know that "!X86 && (X86_32 || X86_64)" can happen (with UML).

Now you could say that's the driver bug, or you could say that they should just add "depends on !UML" (though that's basically equivalent to adding "depends on X86" and the latter may be preferable in some cases).

Or actually in the three patches you have (1-3) it's in the code, but same thing, you can either add && !UML (like you did) or add && X86.

Arguably, however, building PCI drivers by default is somewhat questionable in the first place?

So maybe you should just add

# CONFIG_UML_PCI_OVER_VIRTIO is not set

to the broken_on_uml.config since it exposes all these issues, and really is not very useful since you're not going to actually run with any simulated PCI devices anyway, so drivers will not be probed.

johannes

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

author
#9 lore raw

Re: [PATCH 4/4] kunit: tool: Disable broken options for --alltests

Quoted 8 lines · Johannes Berg
Quoted 3 lines · David Gow

Note that, while this does build again, it still segfaults on startup, so more work remains to be done.

That's probably just a lot more stuff getting included somehow?

Yeah: it used to work (a couple of years ago), but something has broken it in the meantime. It's just a shame that bisecting things with allyesconfig takes so long...

Quoted 42 lines
Quoted 8 lines

They are:
- CONFIG_VFIO_PCI: Needs ioport_map/ioport_unmap.
- CONFIG_INFINIBAND_RDMAVT: Needs cpuinfo_x86 and __copy_user_nocache
- CONFIG_BNXT: Failing under UML with -Werror
ERROR:root:../drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c: In function ‘bnxt_ptp_enable’:
../drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c:400:43: error: array subscript 255 is above array bounds of ‘struct pps_pin[4]’ [-Werror=array-bounds]

  400 |                         ptp->pps_info.pins[pin_id].event = BNXT_PPS_EVENT_EXTERNAL;
      |                         ~~~~~~~~~~~~~~~~~~^~~~~~~~
Quoted 9 lines

- CONFIG_PATA_CS5535: Needs MSR access (__tracepoint_{read,write}_msr)
- CONFIG_VDPA: Enables CONFIG_DMA_OPS, which is unimplemented. ('dma_ops' is not defined)

These are all issues which should be investigated properly and the corresponding options either fixed or disabled under UML. Having this list of broken options should act as a good to-do list here, and will allow these issues to be worked on independently, and other tests to work in the meantime.

I'm not really sure it makes sense to even do anything other than disabling these.

It looks like all of them are just exposed by now being able to build PCI drivers on UML. Surely the people writing the driver didn't expect their drivers to run over simulated PCI (which is what the UML PCI support is all about).

Now from a PCI driver point of view you can't really tell the difference (and anyway the driver won't be probed), but the issues (at least the build time ones) come from having

    UML && PCI && X86_64

or

    UML && PCI && X86_32

because drivers typically depend on X86_64 or X86_32, rather than on "X86 && X86_64" or "X86 && X86_32". In a sense thus, the issue is those drivers don't know that "!X86 && (X86_32 || X86_64)" can happen (with UML).

Quoted 7 lines

Now you could say that's the driver bug, or you could say that they should just add "depends on !UML" (though that's basically equivalent to adding "depends on X86" and the latter may be preferable in some cases).

Or actually in the three patches you have (1-3) it's in the code, but same thing, you can either add && !UML (like you did) or add && X86.

I didn't realise X86 wasn't defined in UML: that's definitely a bit cleaner than !UML in a number of these cases.

Not all of those issues are fundamentally solved by "depends on X86",
though: there are a few which might be other missing things in UML (maybe the 'dma_ops' issues?), and/or might be the result of -Werror being enabled.

Quoted 3 lines

Arguably, however, building PCI drivers by default is somewhat questionable in the first place?

We do want the ability to build PCI drivers under UML, as it makes running KUnit tests for PCI drivers much simpler and more pleasant. And indeed, it does work for KUnit in general, it's just that some drivers do have the issues mentioned above, so allyesconfig picks up every broken driver.

We don't actually build the PCI drivers by default, only if the "--alltests" option is passed, which does include them, as we do have tests which depend on PCI we'd like to run (like the thunderbolt test).

Quoted 8 lines

So maybe you should just add

# CONFIG_UML_PCI_OVER_VIRTIO is not set

to the broken_on_uml.config since it exposes all these issues, and really is not very useful since you're not going to actually run with any simulated PCI devices anyway, so drivers will not be probed.

I did try this as well, and it just got us a different set of issues (there are a bunch of drivers which depend on IOMEM but don't state it -- I'll try to send fixes for those out next week). And we'd lose things like the thunderbolt test if PCI

Ultimately, the 'broken_on_uml.config' file is just there to pare back allyesconfig a bit for KUnit's purposes, but we still definitely want as many options (and hence tests) enabled as possible long-term. So I think actual fixes to either the code or Kconfig do make sense.

Is 'make ARCH=um allyesconfig' something we actually want to be able to build? If so, no amount of adding things to KUnit's broken_on_uml.config will solve the underlying issues, and we'll need to at least update the Kconfig entries.

Cheers,
-- David

maintainer
#10 lore raw

Re: [PATCH 4/4] kunit: tool: Disable broken options for --alltests

Quoted 7 lines · David Gow
Quoted 6 lines · Johannes Berg
Quoted 3 lines · David Gow

Note that, while this does build again, it still segfaults on startup, so more work remains to be done.

Quoted 6 lines
Quoted 2 lines

That's probably just a lot more stuff getting included somehow?

Yeah: it used to work (a couple of years ago), but something has broken it in the meantime. It's just a shame that bisecting things with allyesconfig takes so long...

Heh, right.

But I guess you could "Kconfig bisect" first, i.e. see what option breaks it? It might not even help to bisect, if it's just some option getting enabled over time. Or perhaps the kernel is just too big for the address space layout if you have allyesconfig? Though that shouldn't be an issue, I think.

Quoted 1 line

I didn't realise X86 wasn't defined in UML:

X86 is the architecture, X86_{32,64} is kind of a selection for how you want things to be built, and it's thus required for UML on x86, because UML imports stuff from X86.

Quoted 2 lines

that's definitely a bit
cleaner than !UML in a number of these cases.

It looks like some (most?) of them don't really work that way though since they're not really platform specific, they just know only about a handful of platforms that they're compatible with.

Quoted 4 lines

Not all of those issues are fundamentally solved by "depends on X86",
though: there are a few which might be other missing things in UML (maybe the 'dma_ops' issues?), and/or might be the result of -Werror being enabled.

Right.

Quoted 2 lines

We do want the ability to build PCI drivers under UML, as it makes running KUnit tests for PCI drivers much simpler and more pleasant.

OK, fair point. I'm thinking about this area in general also right now for iwlwifi, and obviously we're probably the only user of the virtual PCI code that lets us connect the driver to a simulated device on UML (but the driver doesn't really know) :-)

Quoted 3 lines

And indeed, it does work for KUnit in general, it's just that some drivers do have the issues mentioned above, so allyesconfig picks up every broken driver.

Right.

Quoted 4 lines

We don't actually build the PCI drivers by default, only if the "--alltests" option is passed, which does include them, as we do have tests which depend on PCI we'd like to run (like the thunderbolt test).

Makes sense.

Quoted 5 lines

I did try this as well, and it just got us a different set of issues (there are a bunch of drivers which depend on IOMEM but don't state it -- I'll try to send fixes for those out next week).

Fun.

Quoted 4 lines

Ultimately, the 'broken_on_uml.config' file is just there to pare back allyesconfig a bit for KUnit's purposes, but we still definitely want as many options (and hence tests) enabled as possible long-term. So I think actual fixes to either the code or Kconfig do make sense.

Makes sense.

Quoted 5 lines

Is 'make ARCH=um allyesconfig' something we actually want to be able to build? If so, no amount of adding things to KUnit's broken_on_uml.config will solve the underlying issues, and we'll need to at least update the Kconfig entries.

That's a good point, as long as people are doing allyes/randconfig builds on UML, we probably need to have these fixes anyway rather than disabling something for KUnit specifically.

johannes

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

#11 lore raw

Re: [PATCH 4/4] kunit: tool: Disable broken options for --alltests

In reply to David Gow
Quoted 11 lines · David Gow

There are a number of Kconfig options which break compilation under UML with allyesconfig. As kunit_tool's --alltests option is based on allyesconfig and UML, we need to update the list of broken options to make --alltests build again.

Note that, while this does build again, it still segfaults on startup, so more work remains to be done.

They are:
- CONFIG_VFIO_PCI: Needs ioport_map/ioport_unmap.
- CONFIG_INFINIBAND_RDMAVT: Needs cpuinfo_x86 and __copy_user_nocache

It doesn't make sense to patch qib and then turn this option off, it is required to build qib.

Jason