Files
Angelo G. Del Regno 1ad6cc0c62 Merge: Performance improvements.
This patchset brings some performance improvements and the addition of the LZO-RLE
algorithm to the kernel, also usable in zram (yup, tested, works but LZ4 is still ok for us).

The main performance improvement is for SWAP space: the locking has changed and
the swap cache is now split in 64MB trunks.
This gives us a reduction of the median page fault latency of 375%, from 15uS to 4uS,
and an improvement of 192% on the swap throughput (this includes "virtual" swap
devices, like zRAM!). The real world user experience improvement of this on a mobile
device is seen after a day or two of usage, where it usually starts losing just a little
performance due to the large amount of apps kept open in background: now I cannot
notice any more performance loss and the user experience is now basically the same as
if the phone was in its first 2 hours of boot life.

Other performance improvements include, in short:

    UDP v4/v6: 10% more performance on single RX queue
    Userspace applications will be faster when checking running time of threads
    2-5% improvements on heavy multipliers (yeah, not a lot, but was totally free...)
    Improvements on rare conditions during sparsetruncate of about 0.3% to a
    way more rare around 20% improvement (that's never gonna happen, but there
    is no performance drop anywhere).

Tested on SoMC Tama Akatsuki RoW

This was taken from
Repo:
https://github.com/sonyxperiadev/kernel
PR: 2039 ([2.3.2.r1.4] Performance improvements)
2024-08-13 23:11:51 +05:30

83 lines
2.1 KiB
C

/*
* include/linux/pagevec.h
*
* In many places it is efficient to batch an operation up against multiple
* pages. A pagevec is a multipage container which is used for that.
*/
#ifndef _LINUX_PAGEVEC_H
#define _LINUX_PAGEVEC_H
/* 15 pointers + header align the pagevec structure to a power of two */
#define PAGEVEC_SIZE 15
struct page;
struct address_space;
struct pagevec {
unsigned char nr;
bool cold;
bool drained;
struct page *pages[PAGEVEC_SIZE];
};
void __pagevec_release(struct pagevec *pvec);
void __pagevec_lru_add(struct pagevec *pvec);
unsigned pagevec_lookup_entries(struct pagevec *pvec,
struct address_space *mapping,
pgoff_t start, unsigned nr_entries,
pgoff_t *indices);
void pagevec_remove_exceptionals(struct pagevec *pvec);
unsigned pagevec_lookup(struct pagevec *pvec, struct address_space *mapping,
pgoff_t start, unsigned nr_pages);
unsigned pagevec_lookup_range_tag(struct pagevec *pvec,
struct address_space *mapping, pgoff_t *index, pgoff_t end,
int tag);
unsigned pagevec_lookup_range_nr_tag(struct pagevec *pvec,
struct address_space *mapping, pgoff_t *index, pgoff_t end,
int tag, unsigned max_pages);
static inline unsigned pagevec_lookup_tag(struct pagevec *pvec,
struct address_space *mapping, pgoff_t *index, int tag)
{
return pagevec_lookup_range_tag(pvec, mapping, index, (pgoff_t)-1, tag);
}
static inline void pagevec_init(struct pagevec *pvec, int cold)
{
pvec->nr = 0;
pvec->cold = cold;
pvec->drained = false;
}
static inline void pagevec_reinit(struct pagevec *pvec)
{
pvec->nr = 0;
}
static inline unsigned pagevec_count(struct pagevec *pvec)
{
return pvec->nr;
}
static inline unsigned pagevec_space(struct pagevec *pvec)
{
return PAGEVEC_SIZE - pvec->nr;
}
/*
* Add a page to a pagevec. Returns the number of slots still available.
*/
static inline unsigned pagevec_add(struct pagevec *pvec, struct page *page)
{
pvec->pages[pvec->nr++] = page;
return pagevec_space(pvec);
}
static inline void pagevec_release(struct pagevec *pvec)
{
if (pagevec_count(pvec))
__pagevec_release(pvec);
}
#endif /* _LINUX_PAGEVEC_H */