Re: [RFC PATCH V3] zram:calculate available memory when zram is used

[yong w]

Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> 于2021年6月8日周二 下午5:28写道：
>
> On Mon, Jun 07, 2021 at 08:39:14AM -0700, yongw.pur@xxxxxxxxx wrote:
> > From: wangyong <wang.yong12@xxxxxxxxxx>
> >
> > When zram is used, available+Swap free memory is obviously bigger than we
> > actually can use, because zram can compress memory by compression
> > algorithm and zram compressed data will occupy memory too.
> >
> > So, we can count the compression ratio of zram in the kernel. The space
> > will be saved by zram and other swap device are calculated as follows:
> > zram[swapfree - swapfree * compress ratio] + swapdev[swapfree]
> > We can evaluate the available memory of the whole system as:
> > MemAvailable+zram[swapfree - swapfree * compress ratio]+swapdev[swapfree]
> >
> > Add an entry to the /proc/meminfo file, returns swap will save space.
> > Which name is more appropriate is still under consideration.
> > There are several alternative names: SwapAvailable, SwapSaved,
> > SwapCompressible, Which is better?
> >
> > Adding new entries has little effect on user program, since parsers
> > usually parse by keywords
> >
> > Changes from v2:
> > *Add interface description document
> > *Other mistakes and problems fix
> >
> > Changes from v1:
> > *Use a new interface to return memory savings when using swap devices
> > *Zram add min_compr_ratio attr
>
> These "Changes" need to go below the --- line please.
>
> >
> > Signed-off-by: wangyong <wang.yong12@xxxxxxxxxx>
> > ---
> >  Documentation/admin-guide/blockdev/zram.rst |  6 ++
> >  Documentation/filesystems/proc.rst          |  4 ++
> >  drivers/block/zram/zcomp.h                  |  8 +++
> >  drivers/block/zram/zram_drv.c               | 19 ++++++
> >  drivers/block/zram/zram_drv.h               |  1 +
> >  fs/proc/meminfo.c                           |  1 +
> >  include/linux/swap.h                        | 11 ++++
> >  mm/swapfile.c                               | 95 +++++++++++++++++++++++++++++
> >  mm/vmscan.c                                 |  1 +
> >  9 files changed, 146 insertions(+)
> >
> > diff --git a/Documentation/admin-guide/blockdev/zram.rst b/Documentation/admin-guide/blockdev/zram.rst
> > index 700329d..3b7c4c4 100644
> > --- a/Documentation/admin-guide/blockdev/zram.rst
> > +++ b/Documentation/admin-guide/blockdev/zram.rst
> > @@ -283,6 +283,12 @@ a single line of text and contains the following stats separated by whitespace:
> >               Unit: 4K bytes
> >   ============== =============================================================
> >
> > +File /sys/block/zram<id>/min_compr_ratio
> > +
> > +The min_compr_ratio file represents the min_compr_ratio during zram swapping out.The calculation formula is as follows:
> > +(orig_size * 100) / compr_data_size
> > +
> > +
>
>
> sysfs files need to be documented in Documentation/ABI/ files.  You can
> reference them in other documentation files, but they need to be in the
> ABI/ directory as well.
>
> Also please wrap your lines at the proper length and use a ' ' after a
> '.'

OK, I'll be careful next time.

>
> >  9) Deactivate
> >  =============
> >
> > diff --git a/Documentation/filesystems/proc.rst b/Documentation/filesystems/proc.rst
> > index 042c418..15d35ae 100644
> > --- a/Documentation/filesystems/proc.rst
> > +++ b/Documentation/filesystems/proc.rst
> > @@ -961,6 +961,7 @@ You may not have all of these fields.
> >      LowFree:          4432 kB
> >      SwapTotal:           0 kB
> >      SwapFree:            0 kB
> > +    SwapAvailable:       0 kB
> >      Dirty:             968 kB
> >      Writeback:           0 kB
> >      AnonPages:      861800 kB
> > @@ -1032,6 +1033,9 @@ SwapTotal
> >  SwapFree
> >                Memory which has been evicted from RAM, and is temporarily
> >                on the disk
> > +SwapAvailable
> > +              The memory savings when use swap devices. it takes zram
> > +              compression ratio into considerations, when zram is used
>
> Trailing whitespace?
>
> Did you run your patch through scripts/checkpatch.pl first before
> sending it out?
>

OK, I'll be careful next time.

>
> >  Dirty
> >                Memory which is waiting to get written back to the disk
> >  Writeback
> > diff --git a/drivers/block/zram/zcomp.h b/drivers/block/zram/zcomp.h
> > index 40f6420..9c9cb96 100644
> > --- a/drivers/block/zram/zcomp.h
> > +++ b/drivers/block/zram/zcomp.h
> > @@ -40,4 +40,12 @@ int zcomp_decompress(struct zcomp_strm *zstrm,
> >               const void *src, unsigned int src_len, void *dst);
> >
> >  bool zcomp_set_max_streams(struct zcomp *comp, int num_strm);
> > +#ifdef CONFIG_ZRAM
> > +int get_zram_major(void);
> > +#else
> > +int get_zram_major(void)
> > +{
> > +     return -1;
> > +}
> > +#endif
> >  #endif /* _ZCOMP_H_ */
> > diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
> > index fcaf275..8f527e0 100644
> > --- a/drivers/block/zram/zram_drv.c
> > +++ b/drivers/block/zram/zram_drv.c
> > @@ -59,6 +59,10 @@ static void zram_free_page(struct zram *zram, size_t index);
> >  static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
> >                               u32 index, int offset, struct bio *bio);
> >
> > +int get_zram_major(void)
> > +{
> > +     return zram_major;
>
> Why does anyone need the zram major number?
>

Swapfile.c will use zram major to determine whether it is a zram device.
I plan to change it to internal interface later.

>
> > +}
> >
> >  static int zram_slot_trylock(struct zram *zram, u32 index)
> >  {
> > @@ -1040,6 +1044,19 @@ static ssize_t compact_store(struct device *dev,
> >       return len;
> >  }
> >
> > +static ssize_t min_compr_ratio_show(struct device *dev,
> > +             struct device_attribute *attr, char *buf)
> > +{
> > +     struct zram *zram = dev_to_zram(dev);
> > +     ssize_t ret;
> > +
> > +     down_read(&zram->init_lock);
> > +     ret = scnprintf(buf, PAGE_SIZE, "%d\n", atomic_read(&zram->stats.min_compr_ratio));
> > +     up_read(&zram->init_lock);
>
> You are using an atomic variable _AND_ a read lock?  Are you sure that
> makes sense?
>
> And please use sysfs_emit() for sysfs files.
>

It refers to mm_stat_show functioin.
    down_read(&zram->init_lock);
     ....
     ret = scnprintf(buf, PAGE_SIZE,
                        "%8llu %8llu %8llu %8lu %8ld %8llu %8lu %8llu %8llu\n",
                        ...
                        (u64)atomic64_read(&zram->stats.same_pages),
                        atomic_long_read(&pool_stats.pages_compacted),
                        (u64)atomic64_read(&zram->stats.huge_pages),
                        (u64)atomic64_read(&zram->stats.huge_pages_since));
        up_read(&zram->init_lock);

>
> > +
> > +     return ret;
> > +}
> > +
> >  static ssize_t io_stat_show(struct device *dev,
> >               struct device_attribute *attr, char *buf)
> >  {
> > @@ -1132,6 +1149,7 @@ static ssize_t debug_stat_show(struct device *dev,
> >       return ret;
> >  }
> >
> > +static DEVICE_ATTR_RO(min_compr_ratio);
> >  static DEVICE_ATTR_RO(io_stat);
> >  static DEVICE_ATTR_RO(mm_stat);
> >  #ifdef CONFIG_ZRAM_WRITEBACK
> > @@ -1859,6 +1877,7 @@ static struct attribute *zram_disk_attrs[] = {
> >       &dev_attr_idle.attr,
> >       &dev_attr_max_comp_streams.attr,
> >       &dev_attr_comp_algorithm.attr,
> > +     &dev_attr_min_compr_ratio.attr,
> >  #ifdef CONFIG_ZRAM_WRITEBACK
> >       &dev_attr_backing_dev.attr,
> >       &dev_attr_writeback.attr,
> > diff --git a/drivers/block/zram/zram_drv.h b/drivers/block/zram/zram_drv.h
> > index 80c3b43..5717e06 100644
> > --- a/drivers/block/zram/zram_drv.h
> > +++ b/drivers/block/zram/zram_drv.h
> > @@ -88,6 +88,7 @@ struct zram_stats {
> >       atomic64_t bd_reads;            /* no. of reads from backing device */
> >       atomic64_t bd_writes;           /* no. of writes from backing device */
> >  #endif
> > +     atomic_t min_compr_ratio;
> >  };
> >
> >  struct zram {
> > diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c
> > index 6fa761c..34a174b 100644
> > --- a/fs/proc/meminfo.c
> > +++ b/fs/proc/meminfo.c
> > @@ -86,6 +86,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
> >
> >       show_val_kb(m, "SwapTotal:      ", i.totalswap);
> >       show_val_kb(m, "SwapFree:       ", i.freeswap);
> > +     show_val_kb(m, "SwapAvailable:  ", count_avail_swaps());
> >       show_val_kb(m, "Dirty:          ",
> >                   global_node_page_state(NR_FILE_DIRTY));
> >       show_val_kb(m, "Writeback:      ",
> > diff --git a/include/linux/swap.h b/include/linux/swap.h
> > index bb48893..deed141 100644
> > --- a/include/linux/swap.h
> > +++ b/include/linux/swap.h
> > @@ -515,6 +515,8 @@ extern int init_swap_address_space(unsigned int type, unsigned long nr_pages);
> >  extern void exit_swap_address_space(unsigned int type);
> >  extern struct swap_info_struct *get_swap_device(swp_entry_t entry);
> >  sector_t swap_page_sector(struct page *page);
> > +extern void update_zram_zstats(void);
> > +extern u64 count_avail_swaps(void);
> >
> >  static inline void put_swap_device(struct swap_info_struct *si)
> >  {
> > @@ -689,6 +691,15 @@ static inline swp_entry_t get_swap_page(struct page *page)
> >       return entry;
> >  }
> >
> > +void update_zram_zstats(void)
> > +{
> > +}
> > +
> > +u64 count_avail_swaps(void)
> > +{
> > +     return 0;
> > +}
> > +
> >  #endif /* CONFIG_SWAP */
> >
> >  #ifdef CONFIG_THP_SWAP
> > diff --git a/mm/swapfile.c b/mm/swapfile.c
> > index 1e07d1c..5ce5100 100644
> > --- a/mm/swapfile.c
> > +++ b/mm/swapfile.c
> > @@ -44,6 +44,7 @@
> >  #include <asm/tlbflush.h>
> >  #include <linux/swapops.h>
> >  #include <linux/swap_cgroup.h>
> > +#include "../drivers/block/zram/zram_drv.h"
>
> That's a big hint that this is not correct, please do not do this :(
>
> The core kernel should not depend on a random block driver's code.
>
> thanks,
>
> greg k-h
Yes,  it's a problem.
Using callback function to realize this function is under consideration.

thanks for your reply.