| Message ID | 20180315180030.20001-5-labbott@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 16/03/2018 02:00, Laura Abbott wrote: > The new challenge is to remove VLAs from the kernel > (see https://lkml.org/lkml/2018/3/7/621) > > This patch replaces a VLA with an appropriate call to kmalloc_array. > > Signed-off-by: Laura Abbott <labbott@redhat.com> > --- > v2: Switch to GFP_KERNEL. There was some discussion about if we should > be doing the allocation at all but given a) the allocation is pretty > small and b) we can possibly take a mutex in a called function I think > this is fine. I still think it's a bad idea. It's simple to preallocate the buffer. But it's up to the maintainer. > --- > drivers/gpio/gpio-stmpe.c | 7 ++++++- > 1 file changed, 6 insertions(+), 1 deletion(-) > > diff --git a/drivers/gpio/gpio-stmpe.c b/drivers/gpio/gpio-stmpe.c > index f8d7d1cd8488..c2bb20ace6f5 100644 > --- a/drivers/gpio/gpio-stmpe.c > +++ b/drivers/gpio/gpio-stmpe.c > @@ -369,10 +369,14 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) > struct stmpe *stmpe = stmpe_gpio->stmpe; > u8 statmsbreg; > int num_banks = DIV_ROUND_UP(stmpe->num_gpios, 8); > - u8 status[num_banks]; > + u8 *status; > int ret; > int i; > > + status = kmalloc_array(num_banks, sizeof(*status), GFP_KERNEL); > + if (!status) > + return IRQ_NONE; > + > /* > * the stmpe_block_read() call below, imposes to set statmsbreg > * with the register located at the lowest address. As STMPE1600 > @@ -424,6 +428,7 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) > } > } > > + kfree(status); > return IRQ_HANDLED; > } > >
On 03/18/2018 06:29 PM, Phil Reid wrote: > On 16/03/2018 02:00, Laura Abbott wrote: >> The new challenge is to remove VLAs from the kernel >> (see https://lkml.org/lkml/2018/3/7/621) >> >> This patch replaces a VLA with an appropriate call to kmalloc_array. >> >> Signed-off-by: Laura Abbott <labbott@redhat.com> >> --- >> v2: Switch to GFP_KERNEL. There was some discussion about if we should >> be doing the allocation at all but given a) the allocation is pretty >> small and b) we can possibly take a mutex in a called function I think >> this is fine. > > I still think it's a bad idea. It's simple to preallocate the buffer. > But it's up to the maintainer. > I'd feel a lot more confident about doing the global buffer with guidance from the maintainer. But looking at the platform data, the maximum number of GPIOs is 24, or 3 banks. Maybe we should just always stack allocate the maximum since it's fairly small. Thanks, Laura > >> --- >> drivers/gpio/gpio-stmpe.c | 7 ++++++- >> 1 file changed, 6 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/gpio/gpio-stmpe.c b/drivers/gpio/gpio-stmpe.c >> index f8d7d1cd8488..c2bb20ace6f5 100644 >> --- a/drivers/gpio/gpio-stmpe.c >> +++ b/drivers/gpio/gpio-stmpe.c >> @@ -369,10 +369,14 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) >> struct stmpe *stmpe = stmpe_gpio->stmpe; >> u8 statmsbreg; >> int num_banks = DIV_ROUND_UP(stmpe->num_gpios, 8); >> - u8 status[num_banks]; >> + u8 *status; >> int ret; >> int i; >> + status = kmalloc_array(num_banks, sizeof(*status), GFP_KERNEL); >> + if (!status) >> + return IRQ_NONE; >> + >> /* >> * the stmpe_block_read() call below, imposes to set statmsbreg >> * with the register located at the lowest address. As STMPE1600 >> @@ -424,6 +428,7 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) >> } >> } >> + kfree(status); >> return IRQ_HANDLED; >> } >> > >
On 23/03/2018 05:43, Laura Abbott wrote: > On 03/18/2018 06:29 PM, Phil Reid wrote: >> On 16/03/2018 02:00, Laura Abbott wrote: >>> The new challenge is to remove VLAs from the kernel >>> (see https://lkml.org/lkml/2018/3/7/621) >>> >>> This patch replaces a VLA with an appropriate call to kmalloc_array. >>> >>> Signed-off-by: Laura Abbott <labbott@redhat.com> >>> --- >>> v2: Switch to GFP_KERNEL. There was some discussion about if we should >>> be doing the allocation at all but given a) the allocation is pretty >>> small and b) we can possibly take a mutex in a called function I think >>> this is fine. >> >> I still think it's a bad idea. It's simple to preallocate the buffer. >> But it's up to the maintainer. >> > > I'd feel a lot more confident about doing the global buffer with > guidance from the maintainer. But looking at the platform data, the > maximum number of GPIOs is 24, or 3 banks. Maybe we should just always > stack allocate the maximum since it's fairly small. That's the other way to go. >>> --- >>> drivers/gpio/gpio-stmpe.c | 7 ++++++- >>> 1 file changed, 6 insertions(+), 1 deletion(-) >>> >>> diff --git a/drivers/gpio/gpio-stmpe.c b/drivers/gpio/gpio-stmpe.c >>> index f8d7d1cd8488..c2bb20ace6f5 100644 >>> --- a/drivers/gpio/gpio-stmpe.c >>> +++ b/drivers/gpio/gpio-stmpe.c >>> @@ -369,10 +369,14 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) >>> struct stmpe *stmpe = stmpe_gpio->stmpe; >>> u8 statmsbreg; >>> int num_banks = DIV_ROUND_UP(stmpe->num_gpios, 8); >>> - u8 status[num_banks]; >>> + u8 *status; >>> int ret; >>> int i; >>> + status = kmalloc_array(num_banks, sizeof(*status), GFP_KERNEL); >>> + if (!status) >>> + return IRQ_NONE; >>> + >>> /* >>> * the stmpe_block_read() call below, imposes to set statmsbreg >>> * with the register located at the lowest address. As STMPE1600 >>> @@ -424,6 +428,7 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) >>> } >>> } >>> + kfree(status); >>> return IRQ_HANDLED; >>> } >>> >> >> > > >
Hi Laura, sorry for slow response :/ On Thu, Mar 22, 2018 at 10:43 PM, Laura Abbott <labbott@redhat.com> wrote: > On 03/18/2018 06:29 PM, Phil Reid wrote: >> >> On 16/03/2018 02:00, Laura Abbott wrote: >>> >>> The new challenge is to remove VLAs from the kernel >>> (see https://lkml.org/lkml/2018/3/7/621) >>> >>> This patch replaces a VLA with an appropriate call to kmalloc_array. >>> >>> Signed-off-by: Laura Abbott <labbott@redhat.com> >>> --- >>> v2: Switch to GFP_KERNEL. There was some discussion about if we should >>> be doing the allocation at all but given a) the allocation is pretty >>> small and b) we can possibly take a mutex in a called function I think >>> this is fine. >> >> >> I still think it's a bad idea. It's simple to preallocate the buffer. >> But it's up to the maintainer. >> > > I'd feel a lot more confident about doing the global buffer with > guidance from the maintainer. But looking at the platform data, the > maximum number of GPIOs is 24, or 3 banks. Maybe we should just always > stack allocate the maximum since it's fairly small. Either way works fine, global (in the state container struct stmpe_gpio) or stack allocation for 24 bits. I guess I am maintainer for this, I can test it at least. Yours, Linus Walleij
diff --git a/drivers/gpio/gpio-stmpe.c b/drivers/gpio/gpio-stmpe.c index f8d7d1cd8488..c2bb20ace6f5 100644 --- a/drivers/gpio/gpio-stmpe.c +++ b/drivers/gpio/gpio-stmpe.c @@ -369,10 +369,14 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) struct stmpe *stmpe = stmpe_gpio->stmpe; u8 statmsbreg; int num_banks = DIV_ROUND_UP(stmpe->num_gpios, 8); - u8 status[num_banks]; + u8 *status; int ret; int i; + status = kmalloc_array(num_banks, sizeof(*status), GFP_KERNEL); + if (!status) + return IRQ_NONE; + /* * the stmpe_block_read() call below, imposes to set statmsbreg * with the register located at the lowest address. As STMPE1600 @@ -424,6 +428,7 @@ static irqreturn_t stmpe_gpio_irq(int irq, void *dev) } } + kfree(status); return IRQ_HANDLED; }
The new challenge is to remove VLAs from the kernel (see https://lkml.org/lkml/2018/3/7/621) This patch replaces a VLA with an appropriate call to kmalloc_array. Signed-off-by: Laura Abbott <labbott@redhat.com> --- v2: Switch to GFP_KERNEL. There was some discussion about if we should be doing the allocation at all but given a) the allocation is pretty small and b) we can possibly take a mutex in a called function I think this is fine. --- drivers/gpio/gpio-stmpe.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)