urbit/ares
1
1
Fork 0
mirror of https://github.com/urbit/ares.git synced 2024-11-22 15:08:54 +03:00
Code Issues Packages Projects Releases Wiki Activity

pma: _bt_state_meta_new leave second metapage zeroed

Browse Source 
There's no reason to initialize anything but the first metapage and it's
probably less error prone to leave the second metapage zeroed. On sync, the
first metapage will get memcpy'd to the second, so why partially write some of
the data now?
This commit is contained in:
barter-simsum 2024-04-02 14:35:21 -04:00
parent af2fde6805
commit a6b12d89bd
1 changed files with 2 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
rust/ares_pma/c-src/btree.c
View File

@ -2311,7 +2311,7 @@ _bt_state_read_header(BT_state *state)
static int static int
_bt_state_meta_new(BT_state *state) _bt_state_meta_new(BT_state *state)
{ {
BT_page *p1, *p2; BT_page *p1;
BT_meta meta = {0}; BT_meta meta = {0};
TRACE(); TRACE();
@ -2335,14 +2335,11 @@ _bt_state_meta_new(BT_state *state)
meta.depth = 1; meta.depth = 1;
meta.flags = BP_META; meta.flags = BP_META;
/* initialize the metapages */ /* initialize the first metapage */
p1 = &((BT_page *)state->map)[0]; p1 = &((BT_page *)state->map)[0];
p2 = &((BT_page *)state->map)[1];
/* copy the metadata into the metapages */ /* copy the metadata into the metapages */
memcpy(METADATA(p1), &meta, sizeof meta); memcpy(METADATA(p1), &meta, sizeof meta);
/* ;;: writing to the second metapage really isn't necessary and it's probably better to leave it zeroed */
/* memcpy(METADATA(p2), &meta, sizeof meta); */
/* only the active metapage should be writable (first page) */ /* only the active metapage should be writable (first page) */
if (mprotect(BT_MAPADDR, BT_META_SECTION_WIDTH, BT_PROT_CLEAN) != 0) { if (mprotect(BT_MAPADDR, BT_META_SECTION_WIDTH, BT_PROT_CLEAN) != 0) {