linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2] of: unittest: Disable interrupt node tests for old world MAC systems
@ 2018-09-26  4:06 Guenter Roeck
  2018-09-26  8:04 ` Frank Rowand
  2018-09-27 14:27 ` Rob Herring
  0 siblings, 2 replies; 3+ messages in thread
From: Guenter Roeck @ 2018-09-26  4:06 UTC (permalink / raw)
  To: Frank Rowand; +Cc: Rob Herring, devicetree, linux-kernel, Guenter Roeck

On systems with OF_IMAP_OLDWORLD_MAC set in of_irq_workarounds, the
devicetree interrupt parsing code is different, causing unit tests of
devicetree interrupt nodes to fail. Due to a bug in unittest code, which
tries to dereference an uninitialized pointer, this results in a crash.

OF: /testcase-data/phandle-tests/consumer-a: arguments longer than property
Unable to handle kernel paging request for data at address 0x00bc616e
Faulting instruction address: 0xc08e9468
Oops: Kernel access of bad area, sig: 11 [#1]
BE PREEMPT PowerMac
Modules linked in:
CPU: 0 PID: 1 Comm: swapper Not tainted 4.14.72-rc1-yocto-standard+ #1
task: cf8e0000 task.stack: cf8da000
NIP:  c08e9468 LR: c08ea5bc CTR: c08ea5ac
REGS: cf8dbb50 TRAP: 0300   Not tainted  (4.14.72-rc1-yocto-standard+)
MSR:  00001032 <ME,IR,DR,RI>  CR: 82004044  XER: 00000000
DAR: 00bc616e DSISR: 40000000
GPR00: c08ea5bc cf8dbc00 cf8e0000 c13ca517 c13ca517 c13ca8a0 00000066 00000002
GPR08: 00000063 00bc614e c0b05865 000affff 82004048 00000000 c00047f0 00000000
GPR16: c0a80000 c0a9cc34 c13ca517 c0ad1134 05ffffff 000affff c0b05860 c0abeef8
GPR24: cecec278 cecec278 c0a8c4d0 c0a885e0 c13ca8a0 05ffffff c13ca8a0 c13ca517

NIP [c08e9468] device_node_gen_full_name+0x30/0x15c
LR [c08ea5bc] device_node_string+0x190/0x3c8
Call Trace:
[cf8dbc00] [c007f670] trace_hardirqs_on_caller+0x118/0x1fc (unreliable)
[cf8dbc40] [c08ea5bc] device_node_string+0x190/0x3c8
[cf8dbcb0] [c08eb794] pointer+0x25c/0x4d0
[cf8dbd00] [c08ebcbc] vsnprintf+0x2b4/0x5ec
[cf8dbd60] [c08ec00c] vscnprintf+0x18/0x48
[cf8dbd70] [c008e268] vprintk_store+0x4c/0x22c
[cf8dbda0] [c008ecac] vprintk_emit+0x94/0x130
[cf8dbdd0] [c008ff54] printk+0x5c/0x6c
[cf8dbe10] [c0b8ddd4] of_unittest+0x2220/0x26f8
[cf8dbea0] [c0004434] do_one_initcall+0x4c/0x184
[cf8dbf00] [c0b4534c] kernel_init_freeable+0x13c/0x1d8
[cf8dbf30] [c0004814] kernel_init+0x24/0x118
[cf8dbf40] [c0013398] ret_from_kernel_thread+0x5c/0x64

The problem was observed when running a qemu test for the g3beige machine
with devicetree unittests enabled.

Disable interrupt node tests on affected systems to avoid both false
unittest failures and the crash.

With this patch in place, unittest on the affected system passes with
the following message.

	dt-test ### end of unittest - 144 passed, 0 failed

Fixes: 53a42093d96ef ("of: Add device tree selftests")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
---
v2: Do not use goto to skip tests.
    Provide test log in commit message.

 drivers/of/unittest.c | 26 ++++++++++++++++++--------
 1 file changed, 18 insertions(+), 8 deletions(-)

diff --git a/drivers/of/unittest.c b/drivers/of/unittest.c
index 722537e14848..41b49716ac75 100644
--- a/drivers/of/unittest.c
+++ b/drivers/of/unittest.c
@@ -771,6 +771,9 @@ static void __init of_unittest_parse_interrupts(void)
 	struct of_phandle_args args;
 	int i, rc;
 
+	if (of_irq_workarounds & OF_IMAP_OLDWORLD_MAC)
+		return;
+
 	np = of_find_node_by_path("/testcase-data/interrupts/interrupts0");
 	if (!np) {
 		pr_err("missing testcase data\n");
@@ -845,6 +848,9 @@ static void __init of_unittest_parse_interrupts_extended(void)
 	struct of_phandle_args args;
 	int i, rc;
 
+	if (of_irq_workarounds & OF_IMAP_OLDWORLD_MAC)
+		return;
+
 	np = of_find_node_by_path("/testcase-data/interrupts/interrupts-extended0");
 	if (!np) {
 		pr_err("missing testcase data\n");
@@ -1001,15 +1007,19 @@ static void __init of_unittest_platform_populate(void)
 	pdev = of_find_device_by_node(np);
 	unittest(pdev, "device 1 creation failed\n");
 
-	irq = platform_get_irq(pdev, 0);
-	unittest(irq == -EPROBE_DEFER, "device deferred probe failed - %d\n", irq);
+	if (!(of_irq_workarounds & OF_IMAP_OLDWORLD_MAC)) {
+		irq = platform_get_irq(pdev, 0);
+		unittest(irq == -EPROBE_DEFER,
+			 "device deferred probe failed - %d\n", irq);
 
-	/* Test that a parsing failure does not return -EPROBE_DEFER */
-	np = of_find_node_by_path("/testcase-data/testcase-device2");
-	pdev = of_find_device_by_node(np);
-	unittest(pdev, "device 2 creation failed\n");
-	irq = platform_get_irq(pdev, 0);
-	unittest(irq < 0 && irq != -EPROBE_DEFER, "device parsing error failed - %d\n", irq);
+		/* Test that a parsing failure does not return -EPROBE_DEFER */
+		np = of_find_node_by_path("/testcase-data/testcase-device2");
+		pdev = of_find_device_by_node(np);
+		unittest(pdev, "device 2 creation failed\n");
+		irq = platform_get_irq(pdev, 0);
+		unittest(irq < 0 && irq != -EPROBE_DEFER,
+			 "device parsing error failed - %d\n", irq);
+	}
 
 	np = of_find_node_by_path("/testcase-data/platform-tests");
 	unittest(np, "No testcase data in device tree\n");
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH v2] of: unittest: Disable interrupt node tests for old world MAC systems
  2018-09-26  4:06 [PATCH v2] of: unittest: Disable interrupt node tests for old world MAC systems Guenter Roeck
@ 2018-09-26  8:04 ` Frank Rowand
  2018-09-27 14:27 ` Rob Herring
  1 sibling, 0 replies; 3+ messages in thread
From: Frank Rowand @ 2018-09-26  8:04 UTC (permalink / raw)
  To: Guenter Roeck; +Cc: Rob Herring, devicetree, linux-kernel

On 09/25/18 21:06, Guenter Roeck wrote:
> On systems with OF_IMAP_OLDWORLD_MAC set in of_irq_workarounds, the
> devicetree interrupt parsing code is different, causing unit tests of
> devicetree interrupt nodes to fail. Due to a bug in unittest code, which
> tries to dereference an uninitialized pointer, this results in a crash.
> 
> OF: /testcase-data/phandle-tests/consumer-a: arguments longer than property
> Unable to handle kernel paging request for data at address 0x00bc616e
> Faulting instruction address: 0xc08e9468
> Oops: Kernel access of bad area, sig: 11 [#1]
> BE PREEMPT PowerMac
> Modules linked in:
> CPU: 0 PID: 1 Comm: swapper Not tainted 4.14.72-rc1-yocto-standard+ #1
> task: cf8e0000 task.stack: cf8da000
> NIP:  c08e9468 LR: c08ea5bc CTR: c08ea5ac
> REGS: cf8dbb50 TRAP: 0300   Not tainted  (4.14.72-rc1-yocto-standard+)
> MSR:  00001032 <ME,IR,DR,RI>  CR: 82004044  XER: 00000000
> DAR: 00bc616e DSISR: 40000000
> GPR00: c08ea5bc cf8dbc00 cf8e0000 c13ca517 c13ca517 c13ca8a0 00000066 00000002
> GPR08: 00000063 00bc614e c0b05865 000affff 82004048 00000000 c00047f0 00000000
> GPR16: c0a80000 c0a9cc34 c13ca517 c0ad1134 05ffffff 000affff c0b05860 c0abeef8
> GPR24: cecec278 cecec278 c0a8c4d0 c0a885e0 c13ca8a0 05ffffff c13ca8a0 c13ca517
> 
> NIP [c08e9468] device_node_gen_full_name+0x30/0x15c
> LR [c08ea5bc] device_node_string+0x190/0x3c8
> Call Trace:
> [cf8dbc00] [c007f670] trace_hardirqs_on_caller+0x118/0x1fc (unreliable)
> [cf8dbc40] [c08ea5bc] device_node_string+0x190/0x3c8
> [cf8dbcb0] [c08eb794] pointer+0x25c/0x4d0
> [cf8dbd00] [c08ebcbc] vsnprintf+0x2b4/0x5ec
> [cf8dbd60] [c08ec00c] vscnprintf+0x18/0x48
> [cf8dbd70] [c008e268] vprintk_store+0x4c/0x22c
> [cf8dbda0] [c008ecac] vprintk_emit+0x94/0x130
> [cf8dbdd0] [c008ff54] printk+0x5c/0x6c
> [cf8dbe10] [c0b8ddd4] of_unittest+0x2220/0x26f8
> [cf8dbea0] [c0004434] do_one_initcall+0x4c/0x184
> [cf8dbf00] [c0b4534c] kernel_init_freeable+0x13c/0x1d8
> [cf8dbf30] [c0004814] kernel_init+0x24/0x118
> [cf8dbf40] [c0013398] ret_from_kernel_thread+0x5c/0x64
> 
> The problem was observed when running a qemu test for the g3beige machine
> with devicetree unittests enabled.
> 
> Disable interrupt node tests on affected systems to avoid both false
> unittest failures and the crash.
> 
> With this patch in place, unittest on the affected system passes with
> the following message.
> 
> 	dt-test ### end of unittest - 144 passed, 0 failed
> 
> Fixes: 53a42093d96ef ("of: Add device tree selftests")
> Signed-off-by: Guenter Roeck <linux@roeck-us.net>
> ---
> v2: Do not use goto to skip tests.
>     Provide test log in commit message.
> 
>  drivers/of/unittest.c | 26 ++++++++++++++++++--------
>  1 file changed, 18 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/of/unittest.c b/drivers/of/unittest.c
> index 722537e14848..41b49716ac75 100644
> --- a/drivers/of/unittest.c
> +++ b/drivers/of/unittest.c
> @@ -771,6 +771,9 @@ static void __init of_unittest_parse_interrupts(void)
>  	struct of_phandle_args args;
>  	int i, rc;
>  
> +	if (of_irq_workarounds & OF_IMAP_OLDWORLD_MAC)
> +		return;
> +
>  	np = of_find_node_by_path("/testcase-data/interrupts/interrupts0");
>  	if (!np) {
>  		pr_err("missing testcase data\n");
> @@ -845,6 +848,9 @@ static void __init of_unittest_parse_interrupts_extended(void)
>  	struct of_phandle_args args;
>  	int i, rc;
>  
> +	if (of_irq_workarounds & OF_IMAP_OLDWORLD_MAC)
> +		return;
> +
>  	np = of_find_node_by_path("/testcase-data/interrupts/interrupts-extended0");
>  	if (!np) {
>  		pr_err("missing testcase data\n");
> @@ -1001,15 +1007,19 @@ static void __init of_unittest_platform_populate(void)
>  	pdev = of_find_device_by_node(np);
>  	unittest(pdev, "device 1 creation failed\n");
>  
> -	irq = platform_get_irq(pdev, 0);
> -	unittest(irq == -EPROBE_DEFER, "device deferred probe failed - %d\n", irq);
> +	if (!(of_irq_workarounds & OF_IMAP_OLDWORLD_MAC)) {
> +		irq = platform_get_irq(pdev, 0);
> +		unittest(irq == -EPROBE_DEFER,
> +			 "device deferred probe failed - %d\n", irq);
>  
> -	/* Test that a parsing failure does not return -EPROBE_DEFER */
> -	np = of_find_node_by_path("/testcase-data/testcase-device2");
> -	pdev = of_find_device_by_node(np);
> -	unittest(pdev, "device 2 creation failed\n");
> -	irq = platform_get_irq(pdev, 0);
> -	unittest(irq < 0 && irq != -EPROBE_DEFER, "device parsing error failed - %d\n", irq);
> +		/* Test that a parsing failure does not return -EPROBE_DEFER */
> +		np = of_find_node_by_path("/testcase-data/testcase-device2");
> +		pdev = of_find_device_by_node(np);
> +		unittest(pdev, "device 2 creation failed\n");
> +		irq = platform_get_irq(pdev, 0);
> +		unittest(irq < 0 && irq != -EPROBE_DEFER,
> +			 "device parsing error failed - %d\n", irq);
> +	}
>  
>  	np = of_find_node_by_path("/testcase-data/platform-tests");
>  	unittest(np, "No testcase data in device tree\n");
> 

Thank you Guenter.

Reviewed-by: Frank Rowand <frank.rowand@sony.com>

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH v2] of: unittest: Disable interrupt node tests for old world MAC systems
  2018-09-26  4:06 [PATCH v2] of: unittest: Disable interrupt node tests for old world MAC systems Guenter Roeck
  2018-09-26  8:04 ` Frank Rowand
@ 2018-09-27 14:27 ` Rob Herring
  1 sibling, 0 replies; 3+ messages in thread
From: Rob Herring @ 2018-09-27 14:27 UTC (permalink / raw)
  To: Guenter Roeck; +Cc: Frank Rowand, devicetree, linux-kernel

On Tue, Sep 25, 2018 at 09:06:24PM -0700, Guenter Roeck wrote:
> On systems with OF_IMAP_OLDWORLD_MAC set in of_irq_workarounds, the
> devicetree interrupt parsing code is different, causing unit tests of
> devicetree interrupt nodes to fail. Due to a bug in unittest code, which
> tries to dereference an uninitialized pointer, this results in a crash.
> 
> OF: /testcase-data/phandle-tests/consumer-a: arguments longer than property
> Unable to handle kernel paging request for data at address 0x00bc616e
> Faulting instruction address: 0xc08e9468
> Oops: Kernel access of bad area, sig: 11 [#1]
> BE PREEMPT PowerMac
> Modules linked in:
> CPU: 0 PID: 1 Comm: swapper Not tainted 4.14.72-rc1-yocto-standard+ #1
> task: cf8e0000 task.stack: cf8da000
> NIP:  c08e9468 LR: c08ea5bc CTR: c08ea5ac
> REGS: cf8dbb50 TRAP: 0300   Not tainted  (4.14.72-rc1-yocto-standard+)
> MSR:  00001032 <ME,IR,DR,RI>  CR: 82004044  XER: 00000000
> DAR: 00bc616e DSISR: 40000000
> GPR00: c08ea5bc cf8dbc00 cf8e0000 c13ca517 c13ca517 c13ca8a0 00000066 00000002
> GPR08: 00000063 00bc614e c0b05865 000affff 82004048 00000000 c00047f0 00000000
> GPR16: c0a80000 c0a9cc34 c13ca517 c0ad1134 05ffffff 000affff c0b05860 c0abeef8
> GPR24: cecec278 cecec278 c0a8c4d0 c0a885e0 c13ca8a0 05ffffff c13ca8a0 c13ca517
> 
> NIP [c08e9468] device_node_gen_full_name+0x30/0x15c
> LR [c08ea5bc] device_node_string+0x190/0x3c8
> Call Trace:
> [cf8dbc00] [c007f670] trace_hardirqs_on_caller+0x118/0x1fc (unreliable)
> [cf8dbc40] [c08ea5bc] device_node_string+0x190/0x3c8
> [cf8dbcb0] [c08eb794] pointer+0x25c/0x4d0
> [cf8dbd00] [c08ebcbc] vsnprintf+0x2b4/0x5ec
> [cf8dbd60] [c08ec00c] vscnprintf+0x18/0x48
> [cf8dbd70] [c008e268] vprintk_store+0x4c/0x22c
> [cf8dbda0] [c008ecac] vprintk_emit+0x94/0x130
> [cf8dbdd0] [c008ff54] printk+0x5c/0x6c
> [cf8dbe10] [c0b8ddd4] of_unittest+0x2220/0x26f8
> [cf8dbea0] [c0004434] do_one_initcall+0x4c/0x184
> [cf8dbf00] [c0b4534c] kernel_init_freeable+0x13c/0x1d8
> [cf8dbf30] [c0004814] kernel_init+0x24/0x118
> [cf8dbf40] [c0013398] ret_from_kernel_thread+0x5c/0x64
> 
> The problem was observed when running a qemu test for the g3beige machine
> with devicetree unittests enabled.
> 
> Disable interrupt node tests on affected systems to avoid both false
> unittest failures and the crash.
> 
> With this patch in place, unittest on the affected system passes with
> the following message.
> 
> 	dt-test ### end of unittest - 144 passed, 0 failed
> 
> Fixes: 53a42093d96ef ("of: Add device tree selftests")
> Signed-off-by: Guenter Roeck <linux@roeck-us.net>
> ---
> v2: Do not use goto to skip tests.
>     Provide test log in commit message.
> 
>  drivers/of/unittest.c | 26 ++++++++++++++++++--------
>  1 file changed, 18 insertions(+), 8 deletions(-)

Applied.

Rob

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2018-09-27 14:27 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-09-26  4:06 [PATCH v2] of: unittest: Disable interrupt node tests for old world MAC systems Guenter Roeck
2018-09-26  8:04 ` Frank Rowand
2018-09-27 14:27 ` Rob Herring

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).