RE: [PATCH v8 1/5] cpumask: introduce cpumap_print_to_buf to support large bitmask and list

[Song Bao Hua (Barry Song)]

> -----Original Message-----
> From: Greg KH [mailto:gregkh@xxxxxxxxxxxxxxxxxxx]
> Sent: Friday, August 6, 2021 12:53 AM
> To: Song Bao Hua (Barry Song) <song.bao.hua@xxxxxxxxxxxxx>
> Cc: andriy.shevchenko@xxxxxxxxxxxxxxx; yury.norov@xxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx; akpm@xxxxxxxxxxxxxxxxxxxx;
> dave.hansen@xxxxxxxxx; linux@xxxxxxxxxxxxxxxxxx; rafael@xxxxxxxxxx;
> rdunlap@xxxxxxxxxxxxx; agordeev@xxxxxxxxxxxxx; sbrivio@xxxxxxxxxx;
> jianpeng.ma@xxxxxxxxx; valentin.schneider@xxxxxxx; peterz@xxxxxxxxxxxxx;
> bristot@xxxxxxxxxx; guodong.xu@xxxxxxxxxx; tangchengchang
> <tangchengchang@xxxxxxxxxx>; Zengtao (B) <prime.zeng@xxxxxxxxxxxxx>;
> yangyicong <yangyicong@xxxxxxxxxx>; tim.c.chen@xxxxxxxxxxxxxxx; Linuxarm
> <linuxarm@xxxxxxxxxx>; tiantao (H) <tiantao6@xxxxxxxxxxxxx>; Jonathan Cameron
> <jonathan.cameron@xxxxxxxxxx>
> Subject: Re: [PATCH v8 1/5] cpumask: introduce cpumap_print_to_buf to support
> large bitmask and list
> 
> On Thu, Jul 29, 2021 at 05:42:04PM +1200, Barry Song wrote:
> > From: Tian Tao <tiantao6@xxxxxxxxxxxxx>
> >
> > The existing cpumap_print_to_pagebuf() is used by cpu topology and other
> > drivers to export hexadecimal bitmask and decimal list to userspace by
> > sysfs ABI.
> >
> > Right now, those drivers are using a normal attribute for this kind of
> > ABIs. A normal attribute typically has show entry as below:
> >
> > static ssize_t example_dev_show(struct device *dev,
> >                 struct device_attribute *attr, char *buf)
> > {
> > 	...
> > 	return cpumap_print_to_pagebuf(true, buf, &pmu_mmdc->cpu);
> > }
> > show entry of attribute has no offset and count parameters and this
> > means the file is limited to one page only.
> >
> > cpumap_print_to_pagebuf() API works terribly well for this kind of
> > normal attribute with buf parameter and without offset, count:
> >
> > static inline ssize_t
> > cpumap_print_to_pagebuf(bool list, char *buf, const struct cpumask *mask)
> > {
> > 	return bitmap_print_to_pagebuf(list, buf, cpumask_bits(mask),
> > 				       nr_cpu_ids);
> > }
> >
> > The problem is once we have many cpus, we have a chance to make bitmask
> > or list more than one page. Especially for list, it could be as complex
> > as 0,3,5,7,9,...... We have no simple way to know it exact size.
> >
> > It turns out bin_attribute is a way to break this limit. bin_attribute
> > has show entry as below:
> > static ssize_t
> > example_bin_attribute_show(struct file *filp, struct kobject *kobj,
> >              struct bin_attribute *attr, char *buf,
> >              loff_t offset, size_t count)
> > {
> > 	...
> > }
> >
> > With the new offset and count parameters, this makes sysfs ABI be able
> > to support file size more than one page. For example, offset could be
> > >= 4096.
> >
> > This patch introduces cpumap_print_to_buf() and its bitmap infrastructure
> > bitmap_print_to_buf() so that those drivers can move to bin_attribute to
> > support large bitmask and list. At the same time, we have to pass those
> > corresponding parameters such as offset, count from bin_attribute to this
> > new API.
> >
> > Signed-off-by: Tian Tao <tiantao6@xxxxxxxxxxxxx>
> > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > Cc: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
> > Cc: Randy Dunlap <rdunlap@xxxxxxxxxxxxx>
> > Cc: Stefano Brivio <sbrivio@xxxxxxxxxx>
> > Cc: Alexander Gordeev <agordeev@xxxxxxxxxxxxx>
> > Cc: "Ma, Jianpeng" <jianpeng.ma@xxxxxxxxx>
> > Cc: Yury Norov <yury.norov@xxxxxxxxx>
> > Cc: Valentin Schneider <valentin.schneider@xxxxxxx>
> > Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > Cc: Daniel Bristot de Oliveira <bristot@xxxxxxxxxx>
> > Reviewed-by: Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx>
> > Signed-off-by: Barry Song <song.bao.hua@xxxxxxxxxxxxx>
> > ---
> >  include/linux/bitmap.h  |  2 ++
> >  include/linux/cpumask.h | 63 +++++++++++++++++++++++++++++++++
> >  lib/bitmap.c            | 78 +++++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 143 insertions(+)
> >
> > diff --git a/include/linux/bitmap.h b/include/linux/bitmap.h
> > index a36cfcec4e77..0de6effa2797 100644
> > --- a/include/linux/bitmap.h
> > +++ b/include/linux/bitmap.h
> > @@ -226,6 +226,8 @@ void bitmap_copy_le(unsigned long *dst, const unsigned
> long *src, unsigned int n
> >  unsigned int bitmap_ord_to_pos(const unsigned long *bitmap, unsigned int ord,
> unsigned int nbits);
> >  int bitmap_print_to_pagebuf(bool list, char *buf,
> >  				   const unsigned long *maskp, int nmaskbits);
> > +int bitmap_print_to_buf(bool list, char *buf, const unsigned long *maskp,
> > +			int nmaskbits, loff_t off, size_t count);
> >
> >  #define BITMAP_FIRST_WORD_MASK(start) (~0UL << ((start) & (BITS_PER_LONG -
> 1)))
> >  #define BITMAP_LAST_WORD_MASK(nbits) (~0UL >> (-(nbits) & (BITS_PER_LONG -
> 1)))
> > diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
> > index f3689a52bfd0..4b838eb6294c 100644
> > --- a/include/linux/cpumask.h
> > +++ b/include/linux/cpumask.h
> > @@ -983,6 +983,69 @@ cpumap_print_to_pagebuf(bool list, char *buf, const
> struct cpumask *mask)
> >  				      nr_cpu_ids);
> >  }
> >
> > +/**
> > + * cpumap_print_to_buf  - copies the cpumask into the buffer
> > + * @list: indicates whether the cpumap must be list
> > + *      true:  print in decimal list format
> > + *      false: print in hexadecimal bitmask format
> > + *
> > + * The existing cpumap_print_to_pagebuf() is used by cpu topology and other
> > + * drivers to export hexadecimal bitmask and decimal list to userspace by
> > + * sysfs ABI.
> > + * Drivers might be using a normal attribute for this kind of ABIs. A
> > + * normal attribute typically has show entry as below:
> > + * static ssize_t example_attribute_show(struct device *dev,
> > + * 		struct device_attribute *attr, char *buf)
> > + * {
> > + * 	...
> > + * 	return cpumap_print_to_pagebuf(true, buf, &pmu_mmdc->cpu);
> > + * }
> > + * show entry of attribute has no offset and count parameters. this means
> > + * the file is limited to one page only.
> > + * cpumap_print_to_pagebuf() API works terribly well for this kind of
> > + * normal attribute with buf parameter and without offset, count:
> > + * cpumap_print_to_pagebuf(bool list, char *buf, const struct cpumask *mask)
> > + * {
> > + * }
> > + * The problem is once we have many cpus, we have a chance to make bitmask
> > + * or list more than one page. Especially for list, it could be as complex
> > + * as 0,3,5,7,9,... We have no simple way to know it exact size.
> > + * It turns out bin_attribute is a way to break this limit. bin_attribute
> > + * has show entry as below:
> > + * static ssize_t
> > + * example_bin_attribute_show(struct file *filp, struct kobject *kobj,
> > + * 		struct bin_attribute *attr, char *buf,
> > + * 		loff_t offset, size_t count)
> > + * {
> > + * 	...
> > + * }
> > + * With the new offset and count parameters, this makes sysfs ABI be able
> > + * to support file size more than one page. For example, offset could be
> > + * >= 4096.
> > + * cpumap_print_to_buf() makes those drivers be able to support large
> > + * bitmask and list after they move to use bin_attribute. In result, we
> > + * have to pass the corresponding parameters such as off, count from
> > + * bin_attribute show entry to this API.
> > + *
> > + * @mask: the cpumask to copy
> > + * @buf: the buffer to copy into
> > + * @off: in the string from which we are copying, We copy to @buf
> > + * @count: the maximum number of bytes to print
> > + *
> > + * The function copies the cpumask into the buffer either as comma-separated
> > + * list of cpus or hex values of cpumask; Typically used by bin_attribute
> to
> > + * export cpumask bitmask and list ABI.
> > + *
> > + * Returns the length of how many bytes have been copied.
> > + */
> > +static inline ssize_t
> > +cpumap_print_to_buf(bool list, char *buf, const struct cpumask *mask,
> > +		loff_t off, size_t count)
> > +{
> > +	return bitmap_print_to_buf(list, buf, cpumask_bits(mask),
> > +				   nr_cpu_ids, off, count);
> > +}
> > +
> >  #if NR_CPUS <= BITS_PER_LONG
> >  #define CPU_MASK_ALL							\
> >  (cpumask_t) { {								\
> > diff --git a/lib/bitmap.c b/lib/bitmap.c
> > index 9401d39e4722..56bcffe2fa8c 100644
> > --- a/lib/bitmap.c
> > +++ b/lib/bitmap.c
> > @@ -487,6 +487,84 @@ int bitmap_print_to_pagebuf(bool list, char *buf, const
> unsigned long *maskp,
> >  }
> >  EXPORT_SYMBOL(bitmap_print_to_pagebuf);
> >
> > +/**
> > + * bitmap_print_to_buf  - convert bitmap to list or hex format ASCII string
> > + * @list: indicates whether the bitmap must be list
> > + *      true:  print in decimal list format
> > + *      false: print in hexadecimal bitmask format
> 
> I do not understand "list" or not here.  Having binary values in a
> function makes it almost impossible to read when they are being called
> without having to go dig up the real user.
> 
> Are you using this in both ways?  If so, make this the common, static
> function and have:
> 	bitmap_print_list_to_buf()
> 	bitmap_print_bitmask_to_buf()
> 
> so that the caller knows exactly what is happening here.

Sounds reasonable to remove bool parameter.

> 
> 
> 
> > + *
> > + * The bitmap_print_to_pagebuf() is used indirectly via its cpumap wrapper
> > + * cpumap_print_to_pagebuf() or directly by drivers to export hexadecimal
> > + * bitmask and decimal list to userspace by sysfs ABI.
> > + * Drivers might be using a normal attribute for this kind of ABIs. A
> > + * normal attribute typically has show entry as below:
> > + * static ssize_t example_attribute_show(struct device *dev,
> > + * 		struct device_attribute *attr, char *buf)
> > + * {
> > + * 	...
> > + * 	return bitmap_print_to_pagebuf(true, buf, &mask, nr_trig_max);
> > + * }
> > + * show entry of attribute has no offset and count parameters and this
> > + * means the file is limited to one page only.
> > + * bitmap_print_to_pagebuf() API works terribly well for this kind of
> > + * normal attribute with buf parameter and without offset, count:
> > + * bitmap_print_to_pagebuf(bool list, char *buf, const unsigned long *maskp,
> > + * 			   int nmaskbits)
> > + * {
> > + * }
> > + * The problem is once we have a large bitmap, we have a chance to get a
> > + * bitmask or list more than one page. Especially for list, it could be
> > + * as complex as 0,3,5,7,9,... We have no simple way to know it exact size.
> > + * It turns out bin_attribute is a way to break this limit. bin_attribute
> > + * has show entry as below:
> > + * static ssize_t
> > + * example_bin_attribute_show(struct file *filp, struct kobject *kobj,
> > + * 		struct bin_attribute *attr, char *buf,
> > + * 		loff_t offset, size_t count)
> > + * {
> > + * 	...
> > + * }
> > + * With the new offset and count parameters, this makes sysfs ABI be able
> > + * to support file size more than one page. For example, offset could be
> > + * >= 4096.
> > + * bitmap_print_to_buf() and its cpumap wrapper cpumap_print_to_buf() makes
> > + * those drivers be able to support large bitmask and list after they move
> > + * to use bin_attribute. In result, we have to pass the corresponding
> > + * parameters such as off, count from bin_attribute show entry to this API.
> > + *
> > + * @buf: buffer into which string is placed
> > + * @maskp: pointer to bitmap to convert
> > + * @nmaskbits: size of bitmap, in bits
> > + * @off: in the string from which we are copying, We copy to @buf
> > + * @count: the maximum number of bytes to print
> > + *
> > + * The role of cpumap_print_to_buf() and cpumap_print_to_pagebuf() is
> similar,
> > + * the difference is that bitmap_print_to_pagebuf() mainly serves sysfs
> > + * attribute with the assumption the destination buffer is exactly one page
> > + * and won't be more than one page. cpumap_print_to_buf(), on the other hand,
> > + * mainly serves bin_attribute which doesn't work with exact one page, and
> it
> > + * can break the size limit of converted decimal list and hexadecimal bitmask.
> > + *
> > + * Returns the number of characters actually printed to @buf
> > + */
> > +int bitmap_print_to_buf(bool list, char *buf, const unsigned long *maskp,
> > +		int nmaskbits, loff_t off, size_t count)
> 
> No need to put the kernel doc for both the .h and .c file, only put it
> in one place please (where ever it ties into the kernel documentation)
> 

Actually they are two different modules. One is cpumap, the other one is
bitmap. But they do have some duplicated content.
I'd prefer to remove the duplicated part from cpumap.

> thanks,
> 
> greg k-h

Thanks
Barry