migration/postcopy: unsentmap is not necessary
cover migration/postcopy: unsentmap is not necessary
Three patches to cleanup postcopy:
[1]: split canonicalize bitmap and discard page
[2]: remove unsentmap since it is not necessary
[3]: cleanup the get_queued_page_not_dirty
Wei Yang (3):
migration/postcopy: not necessary to do discard when canonicalizing
bitmap
migration/postcopy: unsentmap is not necessary for postcopy
migration: remove sent parameter in get_queued_page_not_dirty
include/exec/ram_addr.h | 6 --- migration/ram.c | 94 +++++++---------------------------------- migration/trace-events | 2 +- 3 files changed, 16 insertions(+), 86 deletions(-)
Signature · 1 line
2.17.1
PATCH 1/3 migration/postcopy: not necessary to do discard when canonicalizing bitmap
All pages, either partially sent or partially dirty, will be discarded in postcopy_send_discard_bm_ram(), since we update the unsentmap to be unsentmap = unsentmap | dirty in ram_postcopy_send_discard_bitmap().
This is not necessary to do discard when canonicalizing bitmap. And by doing so, we separate the page discard into two individual steps:
* canonicalize bitmap
* discard page
--- migration/ram.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-)
migration/ram.c +1 -13
@@ -2928,7 +2928,7 @@ static int postcopy_each_ram_send_discard(MigrationState *ms) }/** -* postcopy_chunk_hostpages_pass: canocalize bitmap in hostpages +* postcopy_chunk_hostpages_pass: canonicalize bitmap in hostpages * * Helper for postcopy_chunk_hostpages; it's called twice to * canonicalize the two bitmaps, that are similar, but one is
@@ -2991,18 +2991,6 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass, host_ratio); run_start =QEMU_ALIGN_UP (run_start, host_ratio); -/* Tell the destination to discard this page */ -if (unsent_pass || !test_bit (fixup_start_addr, unsentmap)) { -/* For the unsent_pass we: -* discard partially sent pages -* For the !unsent_pass (dirty) we: -* discard partially dirty pages that were sent -* (any partially sent pages were already discarded -* by the previous unsent_pass) -*/ -postcopy_discard_send_range (ms, fixup_start_addr, host_ratio); - } -/* Clean up the bitmap */ for (page = fixup_start_addr; page < fixup_start_addr + host_ratio; page++) {
Signature · 1 line
2.17.1
Re: [PATCH 1/3] migration/postcopy: not necessary to do discard when canonicalizing bitmap
Quoted 11 lines · * Wei Yang ([email protected]) wrote:
All pages, either partially sent or partially dirty, will be discarded in
postcopy_send_discard_bm_ram(), since we update the unsentmap to be unsentmap = unsentmap | dirty inram_postcopy_send_discard_bitmap().This is not necessary to do discard when canonicalizing bitmap. And by doing so, we separate the page discard into two individual steps:
* canonicalize bitmap
* discard pageSigned-off-by Wei Yang <[email protected]>
Yes, I think when I originally wrote it, the set of pages that was discarded was different; I think it was actually the set of !unsent & dirty - i.e. only pages that had been sent and then redirtied; it later got reworked to include unsent pages as well - so this lot can be simplified.
Quoted 39 lines
--- migration/ram.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-)migration/ram.c +1
-13@@ -2928,7 +2928,7 @@ static int postcopy_each_ram_send_discard(MigrationState *ms) }/** -* postcopy_chunk_hostpages_pass: canocalize bitmap in hostpages +* postcopy_chunk_hostpages_pass: canonicalize bitmap in hostpages * * Helper for postcopy_chunk_hostpages; it's called twice to * canonicalize the two bitmaps, that are similar, but one is @@ -2991,18 +2991,6 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass, host_ratio); run_start =QEMU_ALIGN_UP (run_start, host_ratio); -/* Tell the destination to discard this page */ -if (unsent_pass || !test_bit (fixup_start_addr, unsentmap)) { -/* For the unsent_pass we: -* discard partially sent pages -* For the !unsent_pass (dirty) we: -* discard partially dirty pages that were sent -* (any partially sent pages were already discarded -* by the previous unsent_pass) -*/ -postcopy_discard_send_range (ms, fixup_start_addr, host_ratio); - } -/* Clean up the bitmap */ for (page = fixup_start_addr; page < fixup_start_addr + host_ratio; page++) {Signature · 1 line
2.17.1
--
Dr. David Alan Gilbert / [email protected] / Manchester, UK
Re: [PATCH 1/3] migration/postcopy: not necessary to do discard when canonicalizing bitmap
Quoted 11 lines · Dr. David Alan Gilbert
Quoted 10 lines · * Wei Yang ([email protected]) wrote:
All pages, either partially sent or partially dirty, will be discarded in
postcopy_send_discard_bm_ram(), since we update the unsentmap to be unsentmap = unsentmap | dirty inram_postcopy_send_discard_bitmap().This is not necessary to do discard when canonicalizing bitmap. And by doing so, we separate the page discard into two individual steps:
* canonicalize bitmap
* discard page
Quoted 9 lines
Quoted 1 line
Signed-off-by Wei Yang <[email protected]>Yes, I think when I originally wrote it, the set of pages that was discarded was different; I think it was actually the set of !unsent & dirty - i.e. only pages that had been sent and then redirtied; it later got reworked to include unsent pages as well - so this lot can be simplified.
Thanks for your time :-)
Quoted 44 lines
Reviewed-by Dr. David Alan Gilbert <[email protected]>Quoted 39 lines
--- migration/ram.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-)migration/ram.c +1
-13@@ -2928,7 +2928,7 @@ static int postcopy_each_ram_send_discard(MigrationState *ms) }/** -* postcopy_chunk_hostpages_pass: canocalize bitmap in hostpages +* postcopy_chunk_hostpages_pass: canonicalize bitmap in hostpages * * Helper for postcopy_chunk_hostpages; it's called twice to * canonicalize the two bitmaps, that are similar, but one is @@ -2991,18 +2991,6 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass, host_ratio); run_start =QEMU_ALIGN_UP (run_start, host_ratio); -/* Tell the destination to discard this page */ -if (unsent_pass || !test_bit (fixup_start_addr, unsentmap)) { -/* For the unsent_pass we: -* discard partially sent pages -* For the !unsent_pass (dirty) we: -* discard partially dirty pages that were sent -* (any partially sent pages were already discarded -* by the previous unsent_pass) -*/ -postcopy_discard_send_range (ms, fixup_start_addr, host_ratio); - } -/* Clean up the bitmap */ for (page = fixup_start_addr; page < fixup_start_addr + host_ratio; page++) {Signature · 1 line
2.17.1--
Dr. David Alan Gilbert / [email protected] / Manchester, UK
Signature · 2 lines
Wei Yang Help you, Help me
PATCH 2/3 migration/postcopy: unsentmap is not necessary for postcopy
In reply to Wei YangCommit f3f491fcd6dd594ba695 ('Postcopy: Maintain unsentmap') introduced unsentmap to track not yet sent pages.
This is not necessary since:
* unsentmap is a sub-set of bmap before postcopy start
* unsentmap is the summation of bmap and unsentmap after canonicalizing
This patch just removes it.
--- include/exec/ram_addr.h | 6 ---- migration/ram.c | 80 ++++++++--------------------------------- 2 files changed, 14 insertions(+), 72 deletions(-)
include/exec/ram_addr.h +0 -6
@@ -44,12 +44,6 @@ struct RAMBlock {size_t page_size;/* dirty bitmap used during migration */ unsigned long *bmap; -/* bitmap of pages that haven't been sent even once -* only maintained and used in postcopy at the moment -* where it's used to send the dirtymap at the start -* of the postcopy phase -*/ -unsigned long *unsentmap;/* bitmap of already received pages in postcopy */ unsigned long *receivedmap;
migration/ram.c +14 -66
180 lines not on this page · raw message
Signature · 1 line
2.17.1
Re: [PATCH 2/3] migration/postcopy: unsentmap is not necessary for postcopy
Quoted 11 lines · * Wei Yang ([email protected]) wrote:
Commit
f3f491fcd6dd594ba695('Postcopy: Maintain unsentmap') introduced unsentmap to track not yet sent pages.This is not necessary since:
* unsentmap is a sub-set of bmap before postcopy start
* unsentmap is the summation of bmap and unsentmap after canonicalizingThis patch just removes it.
Signed-off-by Wei Yang <[email protected]>
Yes, that's simpler
Quoted 27 lines
--- include/exec/ram_addr.h | 6 ---- migration/ram.c | 80 ++++++++--------------------------------- 2 files changed, 14 insertions(+), 72 deletions(-)include/exec/ram_addr.h +0
-6@@ -44,12 +44,6 @@ struct RAMBlock {size_t page_size;/* dirty bitmap used during migration */ unsigned long *bmap; -/* bitmap of pages that haven't been sent even once -* only maintained and used in postcopy at the moment -* where it's used to send the dirtymap at the start -* of the postcopy phase -*/ -unsigned long *unsentmap;/* bitmap of already received pages in postcopy */ unsigned long *receivedmap;migration/ram.c +0
-0
migration/ram.c static bool get_queued_page(RAMState *rs, PageSearchStatus *pss) · line 2351
Quoted 5 lines
migration/ram.c +1
-1@@ -2348,7 +2348,7 @@ static bool get_queued_page(RAMState *rs, PageSearchStatus *pss) dirty =test_bit (page, block->bmap);if (!dirty) {trace_get_queued_page_not_dirty (block->idstr, (uint64_t )offset, - page,test_bit (page, block->unsentmap)); + page,test_bit (page, block->bmap));
We could actually lose that last trace parameter.
Dave
Quoted 191 lines
migration/ram.c +13
-65}else {trace_get_queued_page (block->idstr, (uint64_t )offset, page); }@@ -2619,10 +2619,6 @@ static int ram_save_host_page(RAMState *rs, PageSearchStatus *pss, } pages += tmppages; -if (pss->block->unsentmap) { -clear_bit (pss->page, pss->block->unsentmap); - } - pss->page++; }while ((pss->page & (pagesize_bits -1 )) &&offset_in_ramblock (pss->block, pss->page <<TARGET_PAGE_BITS ));@@ -2776,8 +2772,6 @@ static void ram_save_cleanup(void *opaque) block->clear_bmap =NULL ;g_free (block->bmap); block->bmap =NULL ; -g_free (block->unsentmap); - block->unsentmap =NULL ; }xbzrle_cleanup ();@@ -2857,8 +2851,6 @@ void ram_postcopy_migrated_memory_release(MigrationState *ms)* Returns zero on success * * Callback from postcopy_each_ram_send_discard for each RAMBlock -* Note: At this point the 'unsentmap' is the processed bitmap combined -* with the dirtymap; so a '1' means it's either dirty or unsent. * * @ms: current migration state * @block: RAMBlock to discard @@ -2867,17 +2859,17 @@ static int postcopy_send_discard_bm_ram(MigrationState *ms, RAMBlock *block) {unsigned long end = block->used_length >>TARGET_PAGE_BITS ;unsigned long current; -unsigned long *unsentmap = block->unsentmap; +unsigned long *bitmap = block->bmap;for (current =0 ; current < end; ) { -unsigned long one =find_next_bit (unsentmap, end, current); +unsigned long one =find_next_bit (bitmap, end, current);unsigned long zero, discard_length;if (one >= end) {break ; } - zero =find_next_zero_bit (unsentmap, end, one +1 ); + zero =find_next_zero_bit (bitmap, end, one +1 );if (zero >= end) { discard_length = end - one;@@ -2938,16 +2930,12 @@ static int postcopy_each_ram_send_discard(MigrationState *ms)* clean, not a mix. This function canonicalizes the bitmaps. * * @ms: current migration state -* @unsent_pass: if true we need to canonicalize partially unsent host pages -* otherwise we need to canonicalize partially dirty host pages * @block: block that contains the page we want to canonicalize */ -static void postcopy_chunk_hostpages_pass (MigrationState *ms,bool unsent_pass, -RAMBlock *block) +static void postcopy_chunk_hostpages_pass (MigrationState *ms,RAMBlock *block) {RAMState *rs = ram_state;unsigned long *bitmap = block->bmap; -unsigned long *unsentmap = block->unsentmap;unsigned int host_ratio = block->page_size /TARGET_PAGE_SIZE ;unsigned long pages = block->used_length >>TARGET_PAGE_BITS ;unsigned long run_start;@@ -2957,13 +2945,8 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass,return ; } -if (unsent_pass) { -/* Find a sent page */ - run_start =find_next_zero_bit (unsentmap, pages,0 ); - }else { -/* Find a dirty page */ - run_start =find_next_bit (bitmap, pages,0 ); - } +/* Find a dirty page */ + run_start =find_next_bit (bitmap, pages,0 );while (run_start < pages) {@@ -2973,11 +2956,7 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass,*/ if (QEMU_IS_ALIGNED (run_start, host_ratio)) {/* Find the end of this run */ -if (unsent_pass) { - run_start =find_next_bit (unsentmap, pages, run_start +1 ); - }else { - run_start =find_next_zero_bit (bitmap, pages, run_start +1 ); - } + run_start =find_next_zero_bit (bitmap, pages, run_start +1 );/* * If the end isn't at the start of a host page, then the * run doesn't finish at the end of a host page @@ -2994,9 +2973,6 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass,/* Clean up the bitmap */ for (page = fixup_start_addr; page < fixup_start_addr + host_ratio; page++) { -/* All pages in this host page are now not sent */ -set_bit (page, unsentmap); -/* * Remark them as dirty, updating the count for any pages * that weren't previously dirty. @@ -3005,13 +2981,8 @@ static void postcopy_chunk_hostpages_pass(MigrationState *ms, bool unsent_pass, } } -if (unsent_pass) { -/* Find the next sent page for the next iteration */ - run_start =find_next_zero_bit (unsentmap, pages, run_start); - }else { -/* Find the next dirty page for the next iteration */ - run_start =find_next_bit (bitmap, pages, run_start); - } +/* Find the next dirty page for the next iteration */ + run_start =find_next_bit (bitmap, pages, run_start); } }@@ -3033,13 +3004,10 @@ static int postcopy_chunk_hostpages(MigrationState *ms, RAMBlock *block) {postcopy_discard_send_init (ms, block->idstr); -/* First pass: Discard all partially sent host pages */ -postcopy_chunk_hostpages_pass (ms, true, block);/* -* Second pass: Ensure that all partially dirty host pages are made -* fully dirty. +* Ensure that all partially dirty host pages are made fully dirty. */ -postcopy_chunk_hostpages_pass (ms, false, block); +postcopy_chunk_hostpages_pass (ms, block);postcopy_discard_send_finish (ms);return 0 ;@@ -3077,19 +3045,6 @@ int ram_postcopy_send_discard_bitmap(MigrationState *ms) rs->last_page =0 ;RAMBLOCK_FOREACH_NOT_IGNORED (block) { -unsigned long pages = block->used_length >>TARGET_PAGE_BITS ; -unsigned long *bitmap = block->bmap; -unsigned long *unsentmap = block->unsentmap; - -if (!unsentmap) { -/* We don't have a safe way to resize the sentmap, so -* if the bitmap was resized it will be NULL at this -* point. -*/ -error_report ("migration ram resized during precopy phase" ); -rcu_read_unlock (); -return -EINVAL ; - }/* Deal with TPS != HPS and huge pages */ ret =postcopy_chunk_hostpages (ms, block);if (ret) {@@ -3097,12 +3052,9 @@ int ram_postcopy_send_discard_bitmap(MigrationState *ms)return ret; } -/* -* Update the unsentmap to be unsentmap = unsentmap | dirty -*/ -bitmap_or (unsentmap, unsentmap, bitmap, pages);#ifdef DEBUG_POSTCOPY -ram_debug_dump_bitmap (unsentmap, true, pages); +ram_debug_dump_bitmap (block->bmap, true, + block->used_length >>TARGET_PAGE_BITS );#endif }trace_ram_postcopy_send_discard_bitmap ();@@ -3270,10 +3222,6 @@ static void ram_list_init_bitmaps(void)bitmap_set (block->bmap,0 , pages); block->clear_bmap_shift = shift; block->clear_bmap =bitmap_new (clear_bmap_size (pages, shift)); -if (migrate_postcopy_ram ()) { - block->unsentmap =bitmap_new (pages); -bitmap_set (block->unsentmap,0 , pages); - } } } }Signature · 1 line
2.17.1
--
Dr. David Alan Gilbert / [email protected] / Manchester, UK
PATCH 3/3 migration: remove sent parameter in get_queued_page_not_dirty
In reply to Wei YangThis is a cleanup for previous removal of unsentmap.
The sent parameter is not necessary now.
--- migration/ram.c | 2 +- migration/trace-events | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-)
migration/ram.c +1 -1
@@ -2348,7 +2348,7 @@ static bool get_queued_page(RAMState *rs, PageSearchStatus *pss) dirty =test_bit (page, block->bmap);if (!dirty) {trace_get_queued_page_not_dirty (block->idstr, (uint64_t )offset, - page,test_bit (page, block->bmap)); + page); }else {trace_get_queued_page (block->idstr, (uint64_t )offset, page); }
migration/trace-events +1 -1
@@ -76,7 +76,7 @@ qemu_file_fclose(void) "" # ram.c get_queued_page(const char *block_name, uint64_t tmp_offset, unsigned long page_abs) "%s/0x%" PRIx64 " page_abs=0x%lx" -get_queued_page_not_dirty(const char *block_name, uint64_t tmp_offset, unsigned long page_abs, int sent) "%s/0x%" PRIx64 " page_abs=0x%lx (sent=%d)" +get_queued_page_not_dirty(const char *block_name, uint64_t tmp_offset, unsigned long page_abs) "%s/0x%" PRIx64 " page_abs=0x%lx" migration_bitmap_sync_start(void) "" migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64 migration_bitmap_clear_dirty(char *str, uint64_t start, uint64_t size, unsigned long page) "rb %s start 0x%"PRIx64" size 0x%"PRIx64" page 0x%lx"
Signature · 1 line
2.17.1
Re: [PATCH 3/3] migration: remove sent parameter in get_queued_page_not_dirty
Quoted 5 lines · * Wei Yang ([email protected]) wrote:
This is a cleanup for previous removal of unsentmap.
The sent parameter is not necessary now.
Signed-off-by Wei Yang <[email protected]>
OK, you did...
Quoted 34 lines
--- migration/ram.c | 2 +- migration/trace-events | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-)migration/ram.c +1
-1@@ -2348,7 +2348,7 @@ static bool get_queued_page(RAMState *rs, PageSearchStatus *pss) dirty =test_bit (page, block->bmap);if (!dirty) {trace_get_queued_page_not_dirty (block->idstr, (uint64_t )offset, - page,test_bit (page, block->bmap)); + page); }else {trace_get_queued_page (block->idstr, (uint64_t )offset, page); }migration/trace-events +1
-1@@ -76,7 +76,7 @@ qemu_file_fclose(void) "" # ram.c get_queued_page(const char *block_name, uint64_t tmp_offset, unsigned long page_abs) "%s/0x%" PRIx64 " page_abs=0x%lx" -get_queued_page_not_dirty(const char *block_name, uint64_t tmp_offset, unsigned long page_abs, int sent) "%s/0x%" PRIx64 " page_abs=0x%lx (sent=%d)" +get_queued_page_not_dirty(const char *block_name, uint64_t tmp_offset, unsigned long page_abs) "%s/0x%" PRIx64 " page_abs=0x%lx" migration_bitmap_sync_start(void) "" migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64 migration_bitmap_clear_dirty(char *str, uint64_t start, uint64_t size, unsigned long page) "rb %s start 0x%"PRIx64" size 0x%"PRIx64" page 0x%lx"Signature · 1 line
2.17.1
--
Dr. David Alan Gilbert / [email protected] / Manchester, UK
Hi, Dave and Juan
Would you mind taking a look at this one?
Quoted 19 lines · Wei Yang
Three patches to cleanup postcopy:
[1]: split canonicalize bitmap and discard page [2]: remove unsentmap since it is not necessary [3]: cleanup the
get_queued_page_not_dirtyWei Yang (3):
migration/postcopy: not necessary to do discard when canonicalizing
bitmap
migration/postcopy: unsentmap is not necessary for postcopy
migration: remove sent parameter inget_queued_page_not_dirtyinclude/exec/ram_addr.h | 6 --- migration/ram.c | 94 +++++++---------------------------------- migration/trace-events | 2 +- 3 files changed, 16 insertions(+), 86 deletions(-)Signature · 1 line
2.17.1
Signature · 2 lines
Wei Yang Help you, Help me
Quoted 3 lines · * Wei Yang ([email protected]) wrote:
Hi, Dave and Juan
Would you mind taking a look at this one?
Yes, I'll have a look, I'm just fighting through a bunch of other stuff.
Dave
Quoted 13 lines · Wei Yang
Quoted 12 lines
Three patches to cleanup postcopy:
[1]: split canonicalize bitmap and discard page [2]: remove unsentmap since it is not necessary [3]: cleanup the
get_queued_page_not_dirtyWei Yang (3):
migration/postcopy: not necessary to do discard when canonicalizing
bitmap
migration/postcopy: unsentmap is not necessary for postcopy
migration: remove sent parameter inget_queued_page_not_dirty
Quoted 11 lines
Quoted 7 lines
include/exec/ram_addr.h | 6 --- migration/ram.c | 94 +++++++---------------------------------- migration/trace-events | 2 +- 3 files changed, 16 insertions(+), 86 deletions(-)Signature · 1 line
2.17.1Signature · 2 lines
Wei Yang Help you, Help me
--
Dr. David Alan Gilbert / [email protected] / Manchester, UK
Quoted 7 lines · Dr. David Alan Gilbert
Quoted 3 lines · * Wei Yang ([email protected]) wrote:
Hi, Dave and Juan
Would you mind taking a look at this one?
Yes, I'll have a look, I'm just fighting through a bunch of other stuff.
Yep, thanks ~ Me too :-)
Have a good day
Quoted 28 lines
Dave
Quoted 13 lines · Wei Yang
Quoted 12 lines
Three patches to cleanup postcopy:
[1]: split canonicalize bitmap and discard page [2]: remove unsentmap since it is not necessary [3]: cleanup the
get_queued_page_not_dirtyWei Yang (3):
migration/postcopy: not necessary to do discard when canonicalizing
bitmap
migration/postcopy: unsentmap is not necessary for postcopy
migration: remove sent parameter inget_queued_page_not_dirtyQuoted 11 lines
Quoted 7 lines
include/exec/ram_addr.h | 6 --- migration/ram.c | 94 +++++++---------------------------------- migration/trace-events | 2 +- 3 files changed, 16 insertions(+), 86 deletions(-)Signature · 1 line
2.17.1Signature · 2 lines
Wei Yang Help you, Help me--
Dr. David Alan Gilbert / [email protected] / Manchester, UK
Signature · 2 lines
Wei Yang Help you, Help me
Quoted 5 lines · * Wei Yang ([email protected]) wrote:
Three patches to cleanup postcopy:
[1]: split canonicalize bitmap and discard page
[2]: remove unsentmap since it is not necessary
[3]: cleanup theget_queued_page_not_dirty
Queued
Quoted 7 lines
Wei Yang (3):
migration/postcopy: not necessary to do discard when canonicalizing
bitmap
migration/postcopy: unsentmap is not necessary for postcopy
migration: remove sent parameter inget_queued_page_not_dirty
Quoted 9 lines
include/exec/ram_addr.h | 6 --- migration/ram.c | 94 +++++++---------------------------------- migration/trace-events | 2 +- 3 files changed, 16 insertions(+), 86 deletions(-)Signature · 1 line
2.17.1
--
Dr. David Alan Gilbert / [email protected] / Manchester, UK