* [PATCH v3 0/3] Make fault injection available for MMC IO
@ 2011-07-20 22:01 Per Forlin
2011-07-20 22:01 ` [PATCH v3 1/3] fault-inject: make fault injection available for modules Per Forlin
` (2 more replies)
0 siblings, 3 replies; 10+ messages in thread
From: Per Forlin @ 2011-07-20 22:01 UTC (permalink / raw)
To: linux-arm-kernel
The first version of this patch is a part of mmc non-blocking v9 patchset:
[PATCH v9 11/12] mmc: core: add random fault injection
change log:
v2 - Resolve build issue in mmc core.c due to multiple init_module by
removing the fault inject module.
- Export fault injection functions to make them available for modules
- Update fault injection documentation on MMC IO
v3 - add function descriptions in core.c
- use export GPL for fault injection functions
Per Forlin (3):
fault-inject: make fault injection available for modules
mmc: core: add random fault injection
fault injection: add documentation on MMC IO fault injection
Documentation/fault-injection/fault-injection.txt | 5 ++
drivers/mmc/core/core.c | 69 +++++++++++++++++++++
drivers/mmc/core/debugfs.c | 5 ++
include/linux/mmc/host.h | 3 +
lib/Kconfig.debug | 11 +++
lib/fault-inject.c | 2 +
6 files changed, 95 insertions(+), 0 deletions(-)
--
1.7.4.1
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 1/3] fault-inject: make fault injection available for modules
2011-07-20 22:01 [PATCH v3 0/3] Make fault injection available for MMC IO Per Forlin
@ 2011-07-20 22:01 ` Per Forlin
2011-07-20 22:02 ` [PATCH v3 2/3] mmc: core: add random fault injection Per Forlin
2011-07-20 22:02 ` [PATCH v3 3/3] fault injection: add documentation on MMC IO " Per Forlin
2 siblings, 0 replies; 10+ messages in thread
From: Per Forlin @ 2011-07-20 22:01 UTC (permalink / raw)
To: linux-arm-kernel
export symbols should_fail() and init_fault_attr_dentries() in order
to make modules use the fault injection functionality
Signed-off-by: Per Forlin <per.forlin@linaro.org>
---
lib/fault-inject.c | 2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/lib/fault-inject.c b/lib/fault-inject.c
index 7e65af7..27783da 100644
--- a/lib/fault-inject.c
+++ b/lib/fault-inject.c
@@ -131,6 +131,7 @@ bool should_fail(struct fault_attr *attr, ssize_t size)
return true;
}
+EXPORT_SYMBOL_GPL(should_fail);
#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
@@ -311,5 +312,6 @@ fail:
cleanup_fault_attr_dentries(attr);
return -ENOMEM;
}
+EXPORT_SYMBOL_GPL(init_fault_attr_dentries);
#endif /* CONFIG_FAULT_INJECTION_DEBUG_FS */
--
1.7.4.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-20 22:01 [PATCH v3 0/3] Make fault injection available for MMC IO Per Forlin
2011-07-20 22:01 ` [PATCH v3 1/3] fault-inject: make fault injection available for modules Per Forlin
@ 2011-07-20 22:02 ` Per Forlin
2011-07-25 15:58 ` Akinobu Mita
2011-07-20 22:02 ` [PATCH v3 3/3] fault injection: add documentation on MMC IO " Per Forlin
2 siblings, 1 reply; 10+ messages in thread
From: Per Forlin @ 2011-07-20 22:02 UTC (permalink / raw)
To: linux-arm-kernel
This adds support to inject data errors after a completed host transfer.
The mmc core will return error even though the host transfer is successful.
This simple fault injection proved to be very useful to test the
non-blocking error handling in the mmc_blk_issue_rw_rq().
Random faults can also test how the host driver handles pre_req()
and post_req() in case of errors.
Signed-off-by: Per Forlin <per.forlin@linaro.org>
---
drivers/mmc/core/core.c | 69 ++++++++++++++++++++++++++++++++++++++++++++
drivers/mmc/core/debugfs.c | 5 +++
include/linux/mmc/host.h | 3 ++
lib/Kconfig.debug | 11 +++++++
4 files changed, 88 insertions(+), 0 deletions(-)
diff --git a/drivers/mmc/core/core.c b/drivers/mmc/core/core.c
index ab36c7b..dac2673 100644
--- a/drivers/mmc/core/core.c
+++ b/drivers/mmc/core/core.c
@@ -24,6 +24,11 @@
#include <linux/regulator/consumer.h>
#include <linux/pm_runtime.h>
+#ifdef CONFIG_FAIL_MMC_REQUEST
+#include <linux/fault-inject.h>
+#include <linux/random.h>
+#endif
+
#include <linux/mmc/card.h>
#include <linux/mmc/host.h>
#include <linux/mmc/mmc.h>
@@ -82,6 +87,66 @@ static void mmc_flush_scheduled_work(void)
flush_workqueue(workqueue);
}
+#ifdef CONFIG_FAIL_MMC_REQUEST
+
+static DECLARE_FAULT_ATTR(fail_mmc_request);
+
+/*
+ * Internal function. Pass the boot param fail_mmc_request to
+ * the setup fault injection attributes routine.
+ */
+static int __init setup_fail_mmc_request(char *str)
+{
+ return setup_fault_attr(&fail_mmc_request, str);
+}
+__setup("fail_mmc_request=", setup_fail_mmc_request);
+
+/*
+ * Internal function. Inject random data errors.
+ * If mmc_data is NULL no errors are injected.
+ */
+static void mmc_should_fail_request(struct mmc_host *host,
+ struct mmc_request *mrq)
+{
+ struct mmc_command *cmd = mrq->cmd;
+ struct mmc_data *data = mrq->data;
+ static const int data_errors[] = {
+ -ETIMEDOUT,
+ -EILSEQ,
+ -EIO,
+ };
+
+ if (!data)
+ return;
+
+ if (cmd->error || data->error || !host->make_it_fail ||
+ !should_fail(&fail_mmc_request, data->blksz * data->blocks))
+ return;
+
+ data->error = data_errors[random32() % ARRAY_SIZE(data_errors)];
+ data->bytes_xfered = (random32() % (data->bytes_xfered >> 9)) << 9;
+}
+
+static int __init fail_mmc_request_debugfs(void)
+{
+ return init_fault_attr_dentries(&fail_mmc_request,
+ "fail_mmc_request");
+}
+
+#else /* CONFIG_FAIL_MMC_REQUEST */
+
+static void mmc_should_fail_request(struct mmc_host *host,
+ struct mmc_request *mrq)
+{
+}
+
+static int __init fail_mmc_request_debugfs(void)
+{
+ return 0;
+}
+#endif /* CONFIG_FAIL_MMC_REQUEST */
+
+
/**
* mmc_request_done - finish processing an MMC request
* @host: MMC host which completed request
@@ -108,6 +173,8 @@ void mmc_request_done(struct mmc_host *host, struct mmc_request *mrq)
cmd->error = 0;
host->ops->request(host, mrq);
} else {
+ mmc_should_fail_request(host, mrq);
+
led_trigger_event(host->led, LED_OFF);
pr_debug("%s: req done (CMD%u): %d: %08x %08x %08x %08x\n",
@@ -2064,6 +2131,8 @@ static int __init mmc_init(void)
if (ret)
goto unregister_host_class;
+ fail_mmc_request_debugfs();
+
return 0;
unregister_host_class:
diff --git a/drivers/mmc/core/debugfs.c b/drivers/mmc/core/debugfs.c
index 998797e..588e76f 100644
--- a/drivers/mmc/core/debugfs.c
+++ b/drivers/mmc/core/debugfs.c
@@ -188,6 +188,11 @@ void mmc_add_host_debugfs(struct mmc_host *host)
root, &host->clk_delay))
goto err_node;
#endif
+#ifdef CONFIG_FAIL_MMC_REQUEST
+ if (!debugfs_create_u8("make-it-fail", S_IRUSR | S_IWUSR,
+ root, &host->make_it_fail))
+ goto err_node;
+#endif
return;
err_node:
diff --git a/include/linux/mmc/host.h b/include/linux/mmc/host.h
index 771455f..250b46d 100644
--- a/include/linux/mmc/host.h
+++ b/include/linux/mmc/host.h
@@ -303,6 +303,9 @@ struct mmc_host {
struct mmc_async_req *areq; /* active async req */
+#ifdef CONFIG_FAIL_MMC_REQUEST
+ u8 make_it_fail;
+#endif
unsigned long private[0] ____cacheline_aligned;
};
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index c768bcd..c2d1423 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -1057,6 +1057,17 @@ config FAIL_IO_TIMEOUT
Only works with drivers that use the generic timeout handling,
for others it wont do anything.
+config FAIL_MMC_REQUEST
+ bool "Fault-injection capability for MMC IO"
+ select DEBUG_FS
+ depends on FAULT_INJECTION && MMC
+ help
+ Provide fault-injection capability for MMC IO.
+ This will make the mmc core return data errors. This is
+ useful to test the error handling in the mmc block device
+ and to test how the mmc host driver handles retries from
+ the block device.
+
config FAULT_INJECTION_DEBUG_FS
bool "Debugfs entries for fault-injection capabilities"
depends on FAULT_INJECTION && SYSFS && DEBUG_FS
--
1.7.4.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v3 3/3] fault injection: add documentation on MMC IO fault injection
2011-07-20 22:01 [PATCH v3 0/3] Make fault injection available for MMC IO Per Forlin
2011-07-20 22:01 ` [PATCH v3 1/3] fault-inject: make fault injection available for modules Per Forlin
2011-07-20 22:02 ` [PATCH v3 2/3] mmc: core: add random fault injection Per Forlin
@ 2011-07-20 22:02 ` Per Forlin
2 siblings, 0 replies; 10+ messages in thread
From: Per Forlin @ 2011-07-20 22:02 UTC (permalink / raw)
To: linux-arm-kernel
Add description on how to enable random fault injection
for MMC IO
Signed-off-by: Per Forlin <per.forlin@linaro.org>
---
Documentation/fault-injection/fault-injection.txt | 5 +++++
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/Documentation/fault-injection/fault-injection.txt b/Documentation/fault-injection/fault-injection.txt
index 7be15e4..27eede4 100644
--- a/Documentation/fault-injection/fault-injection.txt
+++ b/Documentation/fault-injection/fault-injection.txt
@@ -21,6 +21,11 @@ o fail_make_request
/sys/block/<device>/make-it-fail or
/sys/block/<device>/<partition>/make-it-fail. (generic_make_request())
+o fail_mmc_request
+
+ injects MMC data errors on devices permitted by setting
+ /sys/kernel/debug/mmc0/make-it-fail
+
Configure fault-injection capabilities behavior
-----------------------------------------------
--
1.7.4.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-20 22:02 ` [PATCH v3 2/3] mmc: core: add random fault injection Per Forlin
@ 2011-07-25 15:58 ` Akinobu Mita
2011-07-25 19:19 ` Per Forlin
2011-07-26 20:08 ` Per Forlin
0 siblings, 2 replies; 10+ messages in thread
From: Akinobu Mita @ 2011-07-25 15:58 UTC (permalink / raw)
To: linux-arm-kernel
2011/7/21 Per Forlin <per.forlin@linaro.org>:
> This adds support to inject data errors after a completed host transfer.
> The mmc core will return error even though the host transfer is successful.
> This simple fault injection proved to be very useful to test the
> non-blocking error handling in the mmc_blk_issue_rw_rq().
> Random faults can also test how the host driver handles pre_req()
> and post_req() in case of errors.
Looks good to me.
> @@ -82,6 +87,66 @@ static void mmc_flush_scheduled_work(void)
> ? ? ? ?flush_workqueue(workqueue);
> ?}
>
> +#ifdef CONFIG_FAIL_MMC_REQUEST
> +
> +static DECLARE_FAULT_ATTR(fail_mmc_request);
I think the fail_attr should be defined for each mmc_host like make_it_fail
in struct mmc_host and debugfs entries should also be created in a
subdirectory of mmc host debugfs.
And I know that init_fault_attr_dentries() can only create a
subdirectory in debugfs root directory. But I have a patch which
support for creating it in arbitrary directory. Could you take a look
at this? (Note that this patch is based on mmotm and not yet tested)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-fault-injection-support-for-creating-debugfs-entries.patch
Type: text/x-diff
Size: 6870 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20110726/5e2e0b6b/attachment.bin>
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-25 15:58 ` Akinobu Mita
@ 2011-07-25 19:19 ` Per Forlin
2011-07-25 21:20 ` Per Forlin
2011-07-26 20:08 ` Per Forlin
1 sibling, 1 reply; 10+ messages in thread
From: Per Forlin @ 2011-07-25 19:19 UTC (permalink / raw)
To: linux-arm-kernel
On 25 July 2011 17:58, Akinobu Mita <akinobu.mita@gmail.com> wrote:
> 2011/7/21 Per Forlin <per.forlin@linaro.org>:
>> This adds support to inject data errors after a completed host transfer.
>> The mmc core will return error even though the host transfer is successful.
>> This simple fault injection proved to be very useful to test the
>> non-blocking error handling in the mmc_blk_issue_rw_rq().
>> Random faults can also test how the host driver handles pre_req()
>> and post_req() in case of errors.
>
> Looks good to me.
>
Thanks,
>> @@ -82,6 +87,66 @@ static void mmc_flush_scheduled_work(void)
>> ? ? ? ?flush_workqueue(workqueue);
>> ?}
>>
>> +#ifdef CONFIG_FAIL_MMC_REQUEST
>> +
>> +static DECLARE_FAULT_ATTR(fail_mmc_request);
>
> I think the fail_attr should be defined for each mmc_host like make_it_fail
> in struct mmc_host and debugfs entries should also be created in a
> subdirectory of mmc host debugfs.
>
I looked at blk-core.c and used the same code here. Current code
creates the entry under the debugfs root. This means one fail_attr for
all hosts.
I agree, it's more clean to move the fail_attr to the
host-debugfs-entry which require the fail_attr to be stored same way
as make_it_fail.
> And I know that init_fault_attr_dentries() can only create a
> subdirectory in debugfs root directory. ?But I have a patch which
> support for creating it in arbitrary directory. ?Could you take a look
> at this? (Note that this patch is based on mmotm and not yet tested)
>
Thanks, I will check it out.
Regards,
Per
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-25 19:19 ` Per Forlin
@ 2011-07-25 21:20 ` Per Forlin
2011-07-26 1:41 ` Akinobu Mita
0 siblings, 1 reply; 10+ messages in thread
From: Per Forlin @ 2011-07-25 21:20 UTC (permalink / raw)
To: linux-arm-kernel
Hi Akinobu,
On 25 July 2011 21:19, Per Forlin <per.forlin@linaro.org> wrote:
> On 25 July 2011 17:58, Akinobu Mita <akinobu.mita@gmail.com> wrote:
>> 2011/7/21 Per Forlin <per.forlin@linaro.org>:
>>> This adds support to inject data errors after a completed host transfer.
>>> The mmc core will return error even though the host transfer is successful.
>>> This simple fault injection proved to be very useful to test the
>>> non-blocking error handling in the mmc_blk_issue_rw_rq().
>>> Random faults can also test how the host driver handles pre_req()
>>> and post_req() in case of errors.
>>
>> Looks good to me.
>>
> Thanks,
>
>>> @@ -82,6 +87,66 @@ static void mmc_flush_scheduled_work(void)
>>> ? ? ? ?flush_workqueue(workqueue);
>>> ?}
>>>
>>> +#ifdef CONFIG_FAIL_MMC_REQUEST
>>> +
>>> +static DECLARE_FAULT_ATTR(fail_mmc_request);
>>
>> I think the fail_attr should be defined for each mmc_host like make_it_fail
>> in struct mmc_host and debugfs entries should also be created in a
>> subdirectory of mmc host debugfs.
>>
> I looked at blk-core.c and used the same code here. Current code
> creates the entry under the debugfs root. This means one fail_attr for
> all hosts.
> I agree, it's more clean to move the fail_attr to the
> host-debugfs-entry which require the fail_attr to be stored same way
> as make_it_fail.
>
>> And I know that init_fault_attr_dentries() can only create a
>> subdirectory in debugfs root directory. ?But I have a patch which
>> support for creating it in arbitrary directory. ?Could you take a look
>> at this? (Note that this patch is based on mmotm and not yet tested)
>>
I looked at your patch and it raised two questions.
I can't use FAULT_ATTR_INITIALIZER since mmc_host is allocated on the
heap. It looks like setup_fault_attr(attr, str) will fail if str is
NULL. How can I initialise the fault_attrs if not stack allocated?
About the boot param initialisation of fault attr. There can only be
one fault_mmc_request boot param for the entire kernel but there is
one fault_attr per host, and there may be many hosts. It would be
convenient if setup_fault_attrs would take (attr, boot_param_name),
look up boot_param_name and use that otherwise set default values.
Regards,
Per
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-25 21:20 ` Per Forlin
@ 2011-07-26 1:41 ` Akinobu Mita
2011-07-26 20:01 ` Per Forlin
0 siblings, 1 reply; 10+ messages in thread
From: Akinobu Mita @ 2011-07-26 1:41 UTC (permalink / raw)
To: linux-arm-kernel
2011/7/26 Per Forlin <per.forlin@linaro.org>:
>>> And I know that init_fault_attr_dentries() can only create a
>>> subdirectory in debugfs root directory. ?But I have a patch which
>>> support for creating it in arbitrary directory. ?Could you take a look
>>> at this? (Note that this patch is based on mmotm and not yet tested)
>>>
> I looked at your patch and it raised two questions.
> I can't use FAULT_ATTR_INITIALIZER since mmc_host is allocated on the
> heap. It looks like setup_fault_attr(attr, str) will fail if str is
> NULL. How can I initialise the fault_attrs if not stack allocated?
> About the boot param initialisation of fault attr. There can only be
> one fault_mmc_request boot param for the entire kernel but there is
> one fault_attr per host, and there may be many hosts. It would be
> convenient if setup_fault_attrs would take (attr, boot_param_name),
> look up boot_param_name and use that otherwise set default values.
I think you can define one default fail_attr for boot time configuration
and copy it to per-host fail_attr in mmc_add_host_debugfs().
/* pseudo-code */
static DECLARE_FAULT_ATTR(default_mmc_fail_attr);
static int __init setup_fail_mmc_request(char *str)
{
return setup_fault_attr(&default_mmc_fail_attr, str);
}
__setup("fail_mmc_request=", setup_fail_mmc_request);
...
void mmc_add_host_debugfs(struct mmc_host *host)
{
...
#ifdef CONFIG_FAIL_MMC_REQUEST
host->fail_attr = default_mmc_fail_attr;
if (!debugfs_create_fault_attr("fail_mmc_request",
root, &host->fail_attr))
goto err_node;
#endif
...
}
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-26 1:41 ` Akinobu Mita
@ 2011-07-26 20:01 ` Per Forlin
0 siblings, 0 replies; 10+ messages in thread
From: Per Forlin @ 2011-07-26 20:01 UTC (permalink / raw)
To: linux-arm-kernel
On 26 July 2011 03:41, Akinobu Mita <akinobu.mita@gmail.com> wrote:
> 2011/7/26 Per Forlin <per.forlin@linaro.org>:
>>>> And I know that init_fault_attr_dentries() can only create a
>>>> subdirectory in debugfs root directory. ?But I have a patch which
>>>> support for creating it in arbitrary directory. ?Could you take a look
>>>> at this? (Note that this patch is based on mmotm and not yet tested)
>>>>
>> I looked at your patch and it raised two questions.
>> I can't use FAULT_ATTR_INITIALIZER since mmc_host is allocated on the
>> heap. It looks like setup_fault_attr(attr, str) will fail if str is
>> NULL. How can I initialise the fault_attrs if not stack allocated?
>> About the boot param initialisation of fault attr. There can only be
>> one fault_mmc_request boot param for the entire kernel but there is
>> one fault_attr per host, and there may be many hosts. It would be
>> convenient if setup_fault_attrs would take (attr, boot_param_name),
>> look up boot_param_name and use that otherwise set default values.
>
> I think you can define one default fail_attr for boot time configuration
> and copy it to per-host fail_attr in mmc_add_host_debugfs().
>
You're right. This works fine. I'll prepare a new version of my patch.
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 2/3] mmc: core: add random fault injection
2011-07-25 15:58 ` Akinobu Mita
2011-07-25 19:19 ` Per Forlin
@ 2011-07-26 20:08 ` Per Forlin
1 sibling, 0 replies; 10+ messages in thread
From: Per Forlin @ 2011-07-26 20:08 UTC (permalink / raw)
To: linux-arm-kernel
On 25 July 2011 17:58, Akinobu Mita <akinobu.mita@gmail.com> wrote:
> 2011/7/21 Per Forlin <per.forlin@linaro.org>:
>> This adds support to inject data errors after a completed host transfer.
>> The mmc core will return error even though the host transfer is successful.
>> This simple fault injection proved to be very useful to test the
>> non-blocking error handling in the mmc_blk_issue_rw_rq().
>> Random faults can also test how the host driver handles pre_req()
>> and post_req() in case of errors.
>
> Looks good to me.
>
>> @@ -82,6 +87,66 @@ static void mmc_flush_scheduled_work(void)
>> ? ? ? ?flush_workqueue(workqueue);
>> ?}
>>
>> +#ifdef CONFIG_FAIL_MMC_REQUEST
>> +
>> +static DECLARE_FAULT_ATTR(fail_mmc_request);
>
> I think the fail_attr should be defined for each mmc_host like make_it_fail
> in struct mmc_host and debugfs entries should also be created in a
> subdirectory of mmc host debugfs.
>
> And I know that init_fault_attr_dentries() can only create a
> subdirectory in debugfs root directory. ?But I have a patch which
> support for creating it in arbitrary directory. ?Could you take a look
> at this? (Note that this patch is based on mmotm and not yet tested)
>
I tested your patch on a Snowball board. I had two active mmc cards
and did various fault injection configurations on the two cards
together with dd. I did only test MMC IO fault injection and not any
of the other fault injection clients.
Tested-by: Per Forlin <per.forlin@linaro.org>
Regards,
Per
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2011-07-26 20:08 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-07-20 22:01 [PATCH v3 0/3] Make fault injection available for MMC IO Per Forlin
2011-07-20 22:01 ` [PATCH v3 1/3] fault-inject: make fault injection available for modules Per Forlin
2011-07-20 22:02 ` [PATCH v3 2/3] mmc: core: add random fault injection Per Forlin
2011-07-25 15:58 ` Akinobu Mita
2011-07-25 19:19 ` Per Forlin
2011-07-25 21:20 ` Per Forlin
2011-07-26 1:41 ` Akinobu Mita
2011-07-26 20:01 ` Per Forlin
2011-07-26 20:08 ` Per Forlin
2011-07-20 22:02 ` [PATCH v3 3/3] fault injection: add documentation on MMC IO " Per Forlin
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).