Re: [PATCH] mmc: add ifdef around fault_create_debugfs_attr()

[Adrien Schildknecht]

On Mon, 9 Nov 2015 14:33:11 +0100
Ulf Hansson  wrote:
> On 6 November 2015 at 13:15, Adrien Schildknecht
>  wrote:  
> > FAIL_MMC_REQUEST can be used without FAULT_INJECTION_DEBUG_FS.
> > In this case fault_create_debugfs_attr() will always return an
> > error and lead to the deletion of the whole debugfs directory.
> >
> > This patch makes sure that FAULT_INJECTION_DEBUG_FS is enabled
> > before attempting to create the debugfs atttribute.
> >
> > Signed-off-by: Adrien Schildknecht 
> > ---
> >  drivers/mmc/core/debugfs.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/mmc/core/debugfs.c b/drivers/mmc/core/debugfs.c
> > index 154aced..13e842c 100644
> > --- a/drivers/mmc/core/debugfs.c
> > +++ b/drivers/mmc/core/debugfs.c
> > @@ -259,11 +259,13 @@ void mmc_add_host_debugfs(struct mmc_host
> > *host) if (fail_request)
> > setup_fault_attr(_default_attr, fail_request);
> > host->fail_mmc_request = fail_default_attr;
> > +#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
> > if (IS_ERR(fault_create_debugfs_attr("fail_mmc_request",
> >  root,
> >  >fail_mmc_request)))
> > goto err_node;
> >  #endif
> > +#endif
> > return;
> >
> >  err_node:
> > --
> > 2.6.2
> >
> 
> I think you are solving this in the wrong way. Will it really makes
> sense to use FAIL_MMC_REQUEST unless FAULT_INJECTION_DEBUG_FS is set,
> I don't think so.  
The fault attribute can be configured using either the debugfs or the
module parameter "fail_request". It is thus possible to use
FAIL_MMC_REQUEST without FAULT_INJECTION_DEBUG_FS (even if it is less
convenient).

> I suggest you to change FAIL_MMC_REQUEST from depending on
> FAULT_INJECTION to FAULT_INJECTION_DEBUG_FS in /lib/Kconfig.debug.  
The other systems that use the fault injection infrastructure (futex,
alloc_pages, slab...) can be used without the dependency on
FAULT_INJECTION_DEBUG_FS. They also use nested #ifdef to handle this
configuration. I chose the #ifdef approach for the sake of coherence.

-- 
Adrien Schildknecht
--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] mmc: add ifdef around fault_create_debugfs_attr()

[Ulf Hansson]

On 10 November 2015 at 16:04, Adrien Schildknecht
 wrote:
> On Mon, 9 Nov 2015 14:33:11 +0100
> Ulf Hansson  wrote:
>> On 6 November 2015 at 13:15, Adrien Schildknecht
>>  wrote:
>> > FAIL_MMC_REQUEST can be used without FAULT_INJECTION_DEBUG_FS.
>> > In this case fault_create_debugfs_attr() will always return an
>> > error and lead to the deletion of the whole debugfs directory.
>> >
>> > This patch makes sure that FAULT_INJECTION_DEBUG_FS is enabled
>> > before attempting to create the debugfs atttribute.
>> >
>> > Signed-off-by: Adrien Schildknecht 
>> > ---
>> >  drivers/mmc/core/debugfs.c | 2 ++
>> >  1 file changed, 2 insertions(+)
>> >
>> > diff --git a/drivers/mmc/core/debugfs.c b/drivers/mmc/core/debugfs.c
>> > index 154aced..13e842c 100644
>> > --- a/drivers/mmc/core/debugfs.c
>> > +++ b/drivers/mmc/core/debugfs.c
>> > @@ -259,11 +259,13 @@ void mmc_add_host_debugfs(struct mmc_host
>> > *host) if (fail_request)
>> > setup_fault_attr(_default_attr, fail_request);
>> > host->fail_mmc_request = fail_default_attr;
>> > +#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
>> > if (IS_ERR(fault_create_debugfs_attr("fail_mmc_request",
>> >  root,
>> >  >fail_mmc_request)))
>> > goto err_node;
>> >  #endif
>> > +#endif
>> > return;
>> >
>> >  err_node:
>> > --
>> > 2.6.2
>> >
>>
>> I think you are solving this in the wrong way. Will it really makes
>> sense to use FAIL_MMC_REQUEST unless FAULT_INJECTION_DEBUG_FS is set,
>> I don't think so.
> The fault attribute can be configured using either the debugfs or the
> module parameter "fail_request". It is thus possible to use
> FAIL_MMC_REQUEST without FAULT_INJECTION_DEBUG_FS (even if it is less
> convenient).
>
>> I suggest you to change FAIL_MMC_REQUEST from depending on
>> FAULT_INJECTION to FAULT_INJECTION_DEBUG_FS in /lib/Kconfig.debug.
> The other systems that use the fault injection infrastructure (futex,
> alloc_pages, slab...) can be used without the dependency on
> FAULT_INJECTION_DEBUG_FS. They also use nested #ifdef to handle this

There are also those that chosen to depend on
FAULT_INJECTION_DEBUG_FS, such as FAULT_INJECTION_STACKTRACE_FILTER.

> configuration. I chose the #ifdef approach for the sake of coherence.

I would like to keep the amount of #ifdefs around in the code to a
minimal. In this case it doesn't even make sense and as there are no
consistent behaviour to consider, could you please adopt the Kconfig
solution instead!?

Kind regards
Uffe
--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] mmc: add ifdef around fault_create_debugfs_attr()

[Ulf Hansson]

On 6 November 2015 at 13:15, Adrien Schildknecht  wrote:
> FAIL_MMC_REQUEST can be used without FAULT_INJECTION_DEBUG_FS.
> In this case fault_create_debugfs_attr() will always return an error and
> lead to the deletion of the whole debugfs directory.
>
> This patch makes sure that FAULT_INJECTION_DEBUG_FS is enabled before
> attempting to create the debugfs atttribute.
>
> Signed-off-by: Adrien Schildknecht 
> ---
>  drivers/mmc/core/debugfs.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/mmc/core/debugfs.c b/drivers/mmc/core/debugfs.c
> index 154aced..13e842c 100644
> --- a/drivers/mmc/core/debugfs.c
> +++ b/drivers/mmc/core/debugfs.c
> @@ -259,11 +259,13 @@ void mmc_add_host_debugfs(struct mmc_host *host)
> if (fail_request)
> setup_fault_attr(_default_attr, fail_request);
> host->fail_mmc_request = fail_default_attr;
> +#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
> if (IS_ERR(fault_create_debugfs_attr("fail_mmc_request",
>  root,
>  >fail_mmc_request)))
> goto err_node;
>  #endif
> +#endif
> return;
>
>  err_node:
> --
> 2.6.2
>

I think you are solving this in the wrong way. Will it really makes
sense to use FAIL_MMC_REQUEST unless FAULT_INJECTION_DEBUG_FS is set,
I don't think so.

I suggest you to change FAIL_MMC_REQUEST from depending on
FAULT_INJECTION to FAULT_INJECTION_DEBUG_FS in /lib/Kconfig.debug.

Kind regards
Uffe
--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html