Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian Norris> --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) Applied, thanks. Thierry signature.asc Description: PGP signature
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian Norris > --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) Applied, thanks. Thierry signature.asc Description: PGP signature
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian Norris> --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) Applied, thanks. Thierry signature.asc Description: PGP signature
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian Norris > --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) Applied, thanks. Thierry signature.asc Description: PGP signature
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian NorrisReviewed-by: Guenter Roeck > --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/pwm/pwm-cros-ec.c b/drivers/pwm/pwm-cros-ec.c > index 99b9acc1a420..f6ca4e8c6253 100644 > --- a/drivers/pwm/pwm-cros-ec.c > +++ b/drivers/pwm/pwm-cros-ec.c > @@ -38,7 +38,7 @@ static int cros_ec_pwm_set_duty(struct cros_ec_device *ec, > u8 index, u16 duty) > struct { > struct cros_ec_command msg; > struct ec_params_pwm_set_duty params; > - } buf; > + } __packed buf; > struct ec_params_pwm_set_duty *params = > struct cros_ec_command *msg = > > @@ -65,7 +65,7 @@ static int __cros_ec_pwm_get_duty(struct cros_ec_device > *ec, u8 index, > struct ec_params_pwm_get_duty params; > struct ec_response_pwm_get_duty resp; > }; > - } buf; > + } __packed buf; > struct ec_params_pwm_get_duty *params = > struct ec_response_pwm_get_duty *resp = > struct cros_ec_command *msg = > -- > 2.8.0.rc3.226.g39d4020 >
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian Norris Reviewed-by: Guenter Roeck > --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/pwm/pwm-cros-ec.c b/drivers/pwm/pwm-cros-ec.c > index 99b9acc1a420..f6ca4e8c6253 100644 > --- a/drivers/pwm/pwm-cros-ec.c > +++ b/drivers/pwm/pwm-cros-ec.c > @@ -38,7 +38,7 @@ static int cros_ec_pwm_set_duty(struct cros_ec_device *ec, > u8 index, u16 duty) > struct { > struct cros_ec_command msg; > struct ec_params_pwm_set_duty params; > - } buf; > + } __packed buf; > struct ec_params_pwm_set_duty *params = > struct cros_ec_command *msg = > > @@ -65,7 +65,7 @@ static int __cros_ec_pwm_get_duty(struct cros_ec_device > *ec, u8 index, > struct ec_params_pwm_get_duty params; > struct ec_response_pwm_get_duty resp; > }; > - } buf; > + } __packed buf; > struct ec_params_pwm_get_duty *params = > struct ec_response_pwm_get_duty *resp = > struct cros_ec_command *msg = > -- > 2.8.0.rc3.226.g39d4020 >
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian NorrisReviewed-by: Dmitry Torokhov > --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/pwm/pwm-cros-ec.c b/drivers/pwm/pwm-cros-ec.c > index 99b9acc1a420..f6ca4e8c6253 100644 > --- a/drivers/pwm/pwm-cros-ec.c > +++ b/drivers/pwm/pwm-cros-ec.c > @@ -38,7 +38,7 @@ static int cros_ec_pwm_set_duty(struct cros_ec_device *ec, > u8 index, u16 duty) > struct { > struct cros_ec_command msg; > struct ec_params_pwm_set_duty params; > - } buf; > + } __packed buf; > struct ec_params_pwm_set_duty *params = > struct cros_ec_command *msg = > > @@ -65,7 +65,7 @@ static int __cros_ec_pwm_get_duty(struct cros_ec_device > *ec, u8 index, > struct ec_params_pwm_get_duty params; > struct ec_response_pwm_get_duty resp; > }; > - } buf; > + } __packed buf; > struct ec_params_pwm_get_duty *params = > struct ec_response_pwm_get_duty *resp = > struct cros_ec_command *msg = > -- > 2.8.0.rc3.226.g39d4020 > -- Dmitry
Re: [PATCH] pwm: cros_ec: add __packed to prevent padding
On Tue, Jul 26, 2016 at 11:22:13AM -0700, Brian Norris wrote: > While the particular usage in question is likely safe (struct > cros_ec_command is 32-bit aligned, followed by <=32-bit fields), it's > been suggested this is not a great pattern to follow for the general > case -- for example, if we follow a 'struct cros_ec_command' (which is > 32-bit- but not 64-bit-aligned) with a struct that starts with a 64-bit > type (e.g., u64), the compiler may add padding. > > Let's add __packed, to inform the compiler of our true intention -- to > have no padding between these struct elements -- and to future proof for > any refactorings that might occur. > > Signed-off-by: Brian Norris Reviewed-by: Dmitry Torokhov > --- > Hi Thierry, > > I don't know of any bugs directly resolved by this patch, so take it for > v4.8/v4.9 at your discretion. > > drivers/pwm/pwm-cros-ec.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/pwm/pwm-cros-ec.c b/drivers/pwm/pwm-cros-ec.c > index 99b9acc1a420..f6ca4e8c6253 100644 > --- a/drivers/pwm/pwm-cros-ec.c > +++ b/drivers/pwm/pwm-cros-ec.c > @@ -38,7 +38,7 @@ static int cros_ec_pwm_set_duty(struct cros_ec_device *ec, > u8 index, u16 duty) > struct { > struct cros_ec_command msg; > struct ec_params_pwm_set_duty params; > - } buf; > + } __packed buf; > struct ec_params_pwm_set_duty *params = > struct cros_ec_command *msg = > > @@ -65,7 +65,7 @@ static int __cros_ec_pwm_get_duty(struct cros_ec_device > *ec, u8 index, > struct ec_params_pwm_get_duty params; > struct ec_response_pwm_get_duty resp; > }; > - } buf; > + } __packed buf; > struct ec_params_pwm_get_duty *params = > struct ec_response_pwm_get_duty *resp = > struct cros_ec_command *msg = > -- > 2.8.0.rc3.226.g39d4020 > -- Dmitry