ARMV8 Linux kernel Source analysis: __flush_dcache_all ()

Source: Internet
Author: User

1.1

/*

* __flush_dcache_all ()

* Flush the Wholed-cache.

* Corrupted registers:x0-x7, x9-x11

*/

ENTRY (__flush_dcache_all)

Guarantee the order of previous visit instructions

DSB Sy

Read Cache level ID Register

Mrs X0, CLIDR_EL1//Read CLIDR

Take bits[26:24] (level of coherency for the cache hierarchy.)

Cache consistency is required (e.g. with level 3 cache, but level 2 requires consistency)

And X3, x0, #0x7000000//extract loc from CLIDR

Logical right Shift 23 bits, put bits[26:24] to bits[2:0]

LSR X3, X3, #23//left align loc bit field

If you need to do cache consistency at the level of 0, you do not need to flush and jump to the finished mark.

CBZ X3, finished//If LOC is 0, then no need toclean

//x10 Store Cache level, from level0 Cache start doing Flush

      // the following three loops LOOP3 is a Set/way ( x9 ),

//LOOP2 is a Index ( X7 ), LOOP1 is a Cachelevel (x10)

mov x10, #0//start clean at cache level 0

LOOP1:

X10+2 after the right shift is exactly equal to 1, plus x10 itself is exactly equal to 3

The number of executions per loop1,x2+3* is performed to move the X0 (clidr_el1) to the right by 3 bits,

Take down the CType Type fields field of the cache, Clidr_el1 in the form "ARMv8 ARM"

Add x2, x10, X10, LSR #1/

X0 logical right Shift x2 bit, to X1, extract the cache type into the X1, x0 storage: clidr_el1

LSR x1, x0, x2

Hide the high level and take the current cache type only

and X1, X1, #7

/* Determine what type of current cache is:

* No Cache.

* 001 instruction Cache only.

* 010 Data Cache only.

* 011 separate instruction and data caches.

* Unified Cache.

*/

Less than 2 indicates that the data cache does not exist or only icache,

Jump skip execution, greater than or equal to 2 continue execution

CMP x1, #2

b.lt Skip

/*

* save/disable and restore interrupts.

*. macro Save_and_disable_irqs, Olddaif

* Mrs \olddaif,daif

* DISABLE_IRQ

*. endm

*/

Save Daif to X9 Register, turn off interrupts

Save_and_disable_irqs x9//Make CSSELR and CCSIDR access atomic

Select the current cache level to operate, Csselr_el1 Register BIT[3:1] Select the cache level to operate

X10=0 at first execution, select level 0 cache

MSR csselr_el1,x10

ISB is used to synchronize new CSSR and CSIDR registers

Isb

Because "MSR csselr_el1,x10" is executed, the CCSIDR_EL1 is re-read

Mrs X1, CCSIDR_EL1//Read the new CCSIDR

/*

*. macro Restore_irqs, Olddaif

* MSR Daif, \olddaif

. * ENDM

*/

Restore_irqs x9

X1 Storage Ccsidr_el1 content, low three bits is (Log2 (number of bytes in cache line)) –4

// Plus 4 after x2= (Log2 (numberof bytes in cache line))

And x2, x1, #7//Extract the length of the Cachelines

Add x2, x2, #4//Add 4 (line length offset)

MOV x4, #0x3ff

Logical right Shift 3 bits, extract Bits[12:3] (associativityof cache) –1,

//x4 Storage Cache of the the number

And X4, x4, x1, LSR #3//Find maximum number on the the-the-the-

Calculate the number of X4 front 0 and save to X5

CLZX5, x4//Find bit position of the sizeincrement

Extract bits[27:13] bit: (number of sets in cache)-1

mov x7, #0x7fff

//x7 stored in Cache in the Set number

And X7, X7, x1, LSR #13//Extract max number of the the index size

LOOP2:

Back up the X4 value

mov x9, x4//Create working copy of Max Waysize

LOOP3:

Store the required operation to X6

LSL X6, X9, X5

Determine which tier of the operation (X10 specifies which level of cache to operate)

Orr X11, X10, X6//Factor and cache number intox11

Determine which set to operate

LSL X6, X7, x2

Orr X11, X11, X6//Factor index number into X11

What level of cache (10) is stored in the X11, which cache (x9), which set (X7)

DC CISW, x11//Clean & invalidate by Set/way

Number of-1

Subs X9, x9, #1//Decrementthe

B.ge LOOP3

Subs X7, X7, #1//Decrementthe Index

B.ge LOOP2

Skip

Add x10, X10, #2//increment cache number,

Why is add 2 not 1? See LOOP1 Marking Office explanation

CMP X3, x10

B.GT LOOP1

Finished:

mov x10, #0//Swith back-to-cache level 0

MSR Csselr_el1, x10//Select Current cache level INCSSELR

DSB Sy

Isb

Ret

Endproc (__flush_dcache_all)

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: info-contact@alibabacloud.com and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.