* [PATCH] hwmon: (ads7871) Fix ads7871_probe error path @ 2010-08-16 1:31 Axel Lin 2010-08-17 11:07 ` Jean Delvare 0 siblings, 1 reply; 4+ messages in thread From: Axel Lin @ 2010-08-16 1:31 UTC (permalink / raw) To: linux-kernel; +Cc: Paul Thomas, Jean Delvare, lm-sensors We need to call hwmon_device_unregister() if an error is detected after sucessfully register hwmon device. Signed-off-by: Axel Lin <axel.lin@gmail.com> --- drivers/hwmon/ads7871.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/drivers/hwmon/ads7871.c b/drivers/hwmon/ads7871.c index b300a20..303db92 100644 --- a/drivers/hwmon/ads7871.c +++ b/drivers/hwmon/ads7871.c @@ -201,11 +201,13 @@ static int __devinit ads7871_probe(struct spi_device *spi) we need to make sure we really have a chip*/ if (val != ret) { err = -ENODEV; - goto error_remove; + goto error_unregister; } return 0; +error_unregister: + hwmon_device_unregister(pdata->hwmon_dev); error_remove: sysfs_remove_group(&spi->dev.kobj, &ads7871_group); error_free: -- 1.7.2 ^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] hwmon: (ads7871) Fix ads7871_probe error path 2010-08-16 1:31 [PATCH] hwmon: (ads7871) Fix ads7871_probe error path Axel Lin @ 2010-08-17 11:07 ` Jean Delvare 2010-08-17 15:15 ` Paul Thomas 0 siblings, 1 reply; 4+ messages in thread From: Jean Delvare @ 2010-08-17 11:07 UTC (permalink / raw) To: Axel Lin; +Cc: linux-kernel, Paul Thomas, lm-sensors Hi Alex, On Mon, 16 Aug 2010 09:31:17 +0800, Axel Lin wrote: > We need to call hwmon_device_unregister() if an error is detected after > sucessfully register hwmon device. > > Signed-off-by: Axel Lin <axel.lin@gmail.com> > --- > drivers/hwmon/ads7871.c | 4 +++- > 1 files changed, 3 insertions(+), 1 deletions(-) > > diff --git a/drivers/hwmon/ads7871.c b/drivers/hwmon/ads7871.c > index b300a20..303db92 100644 > --- a/drivers/hwmon/ads7871.c > +++ b/drivers/hwmon/ads7871.c > @@ -201,11 +201,13 @@ static int __devinit ads7871_probe(struct spi_device *spi) > we need to make sure we really have a chip*/ > if (val != ret) { > err = -ENODEV; > - goto error_remove; > + goto error_unregister; > } > > return 0; > > +error_unregister: > + hwmon_device_unregister(pdata->hwmon_dev); > error_remove: > sysfs_remove_group(&spi->dev.kobj, &ads7871_group); > error_free: The bug is real, but I don't think the fix is correct. You should never have to unregister the hwmon device in the error path, because you should ensure the hardware is there and working _before_ you register the hwmon device. User-space could watch for hwmon devices being added or removed, and you don't want to trigger such events for a device which isn't going to work. So I would suggest reworking the order in which things are done, leaving hwmon_device_register() at the end of the function. This won't only fix the error path, this will also close a race window, as for the moment, the device is exposed to user-space _before_ it it properly initialized, which is wrong. Please send an updated patch and I'll be happy to apply it. -- Jean Delvare ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] hwmon: (ads7871) Fix ads7871_probe error path 2010-08-17 11:07 ` Jean Delvare @ 2010-08-17 15:15 ` Paul Thomas 2010-08-17 18:42 ` Jean Delvare 0 siblings, 1 reply; 4+ messages in thread From: Paul Thomas @ 2010-08-17 15:15 UTC (permalink / raw) To: Jean Delvare; +Cc: Axel Lin, linux-kernel, lm-sensors On Tue, Aug 17, 2010 at 4:07 AM, Jean Delvare <khali@linux-fr.org> wrote: > Hi Alex, > > On Mon, 16 Aug 2010 09:31:17 +0800, Axel Lin wrote: >> We need to call hwmon_device_unregister() if an error is detected after >> sucessfully register hwmon device. >> >> Signed-off-by: Axel Lin <axel.lin@gmail.com> >> --- >> drivers/hwmon/ads7871.c | 4 +++- >> 1 files changed, 3 insertions(+), 1 deletions(-) >> >> diff --git a/drivers/hwmon/ads7871.c b/drivers/hwmon/ads7871.c >> index b300a20..303db92 100644 >> --- a/drivers/hwmon/ads7871.c >> +++ b/drivers/hwmon/ads7871.c >> @@ -201,11 +201,13 @@ static int __devinit ads7871_probe(struct spi_device *spi) >> we need to make sure we really have a chip*/ >> if (val != ret) { >> err = -ENODEV; >> - goto error_remove; >> + goto error_unregister; >> } >> >> return 0; >> >> +error_unregister: >> + hwmon_device_unregister(pdata->hwmon_dev); >> error_remove: >> sysfs_remove_group(&spi->dev.kobj, &ads7871_group); >> error_free: > > The bug is real, but I don't think the fix is correct. You should never > have to unregister the hwmon device in the error path, because you > should ensure the hardware is there and working _before_ you register > the hwmon device. > > User-space could watch for hwmon devices being added or removed, and > you don't want to trigger such events for a device which isn't going to > work. > > So I would suggest reworking the order in which things are done, > leaving hwmon_device_register() at the end of the function. This won't > only fix the error path, this will also close a race window, as for the > moment, the device is exposed to user-space _before_ it it properly > initialized, which is wrong. > > Please send an updated patch and I'll be happy to apply it. > > -- > Jean Delvare > Should the kzalloc, sysfs_create_group and hwmon_device_register all stay in the same order just following the register check? I can send a patch for that. thanks, Paul ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] hwmon: (ads7871) Fix ads7871_probe error path 2010-08-17 15:15 ` Paul Thomas @ 2010-08-17 18:42 ` Jean Delvare 0 siblings, 0 replies; 4+ messages in thread From: Jean Delvare @ 2010-08-17 18:42 UTC (permalink / raw) To: Paul Thomas; +Cc: Axel Lin, linux-kernel, lm-sensors On Tue, 17 Aug 2010 08:15:59 -0700, Paul Thomas wrote: > On Tue, Aug 17, 2010 at 4:07 AM, Jean Delvare <khali@linux-fr.org> wrote: > > Hi Alex, > > > > On Mon, 16 Aug 2010 09:31:17 +0800, Axel Lin wrote: > >> We need to call hwmon_device_unregister() if an error is detected after > >> sucessfully register hwmon device. > >> > >> Signed-off-by: Axel Lin <axel.lin@gmail.com> > >> --- > >> drivers/hwmon/ads7871.c | 4 +++- > >> 1 files changed, 3 insertions(+), 1 deletions(-) > >> > >> diff --git a/drivers/hwmon/ads7871.c b/drivers/hwmon/ads7871.c > >> index b300a20..303db92 100644 > >> --- a/drivers/hwmon/ads7871.c > >> +++ b/drivers/hwmon/ads7871.c > >> @@ -201,11 +201,13 @@ static int __devinit ads7871_probe(struct spi_device *spi) > >> we need to make sure we really have a chip*/ > >> if (val != ret) { > >> err = -ENODEV; > >> - goto error_remove; > >> + goto error_unregister; > >> } > >> > >> return 0; > >> > >> +error_unregister: > >> + hwmon_device_unregister(pdata->hwmon_dev); > >> error_remove: > >> sysfs_remove_group(&spi->dev.kobj, &ads7871_group); > >> error_free: > > > > The bug is real, but I don't think the fix is correct. You should never > > have to unregister the hwmon device in the error path, because you > > should ensure the hardware is there and working _before_ you register > > the hwmon device. > > > > User-space could watch for hwmon devices being added or removed, and > > you don't want to trigger such events for a device which isn't going to > > work. > > > > So I would suggest reworking the order in which things are done, > > leaving hwmon_device_register() at the end of the function. This won't > > only fix the error path, this will also close a race window, as for the > > moment, the device is exposed to user-space _before_ it it properly > > initialized, which is wrong. > > > > Please send an updated patch and I'll be happy to apply it. > > > > -- > > Jean Delvare > > > > Should the kzalloc, sysfs_create_group and hwmon_device_register all > stay in the same order just following the register check? I can send a > patch for that. Axel already sent an updated patch, so it's all OK. -- Jean Delvare ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2010-08-17 18:43 UTC | newest] Thread overview: 4+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2010-08-16 1:31 [PATCH] hwmon: (ads7871) Fix ads7871_probe error path Axel Lin 2010-08-17 11:07 ` Jean Delvare 2010-08-17 15:15 ` Paul Thomas 2010-08-17 18:42 ` Jean Delvare
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox