Replaced inlined lfsr_data_t with a lazily encoded leb128

The idea is that we can save on the cost of calling lfs_toleb128
everywhere we commit leb128s, by lazily encoding during progdata.

I original thought this would have too many small problems, but:

1. We can actually implement slice surprisingly easily by just shifting
   the internal word 7 bits. This emulates byte-level slicing in the
   encoded leb128.

   This enables read/cmp, so we can implement all of the lfsr_data_t
   functions, though it does make lfs_toleb128 required for a readonly
   implementation, which isn't great. Sufficient creativity with ifdefs
   likely makes this a non-problem though.

2. There's really very limited use cases for non-leb128 inlined datas.

   We can use it to encode the version and compatflags during
   lfs_format, but that's about it. And lfs_format is definitely not on
   the stack hot-path, so there's no reason to not use on-stack buffers
   for these.

The original motivation for this change was noticing a surprising amount
of code savings related to lazy leb128 encoding in another lfsr_data_t
refactor. Unfortunately this savings does not seem reproducible:

           code          stack
  before: 33864           2880
  after:  33912 (+0.1%)   2888 (+0.3%)

But that's ok, this is closer to what I expected. The lfs_sizeleb128
call we need to predict the leb128 size is close to the same cost as
calling lfs_toleb128 so the savings isn't really that much.
This commit is contained in:
Christopher Haster
2024-02-25 12:08:26 -06:00
parent 5005db2b4e
commit 415e148f62
4 changed files with 86 additions and 97 deletions
+23 -23
View File
@@ -3251,7 +3251,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("aaa", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("aaa", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("0", 1)))) => 0;
printf("btree: w%d 0x%x.%x\n",
btree.weight,
@@ -3304,13 +3304,13 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("aaa", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("aaa", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("0", 1)))) => 0;
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("0", 1)),
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("aab", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("aab", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("1", 1)))) => 0;
printf("btree: w%d 0x%x.%x\n",
btree.weight,
@@ -3372,7 +3372,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF("aaa", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("0", 1)))) => 0;
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
@@ -3380,7 +3380,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(1*DID),
LFSR_DATA_WORD(1*DID),
LFSR_DATA_BUF("aab", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("1", 1)))) => 0;
lfsr_btree_commit(&lfs, &btree, 1, LFSR_ATTRS(
@@ -3388,7 +3388,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(2*DID),
LFSR_DATA_WORD(2*DID),
LFSR_DATA_BUF("aac", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("2", 1)))) => 0;
printf("btree: w%d 0x%x.%x\n",
@@ -3459,7 +3459,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF("aaa", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("0", 1)))) => 0;
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
@@ -3467,7 +3467,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(2*DID),
LFSR_DATA_WORD(2*DID),
LFSR_DATA_BUF("aac", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("2", 1)))) => 0;
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
@@ -3475,7 +3475,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(1*DID),
LFSR_DATA_WORD(1*DID),
LFSR_DATA_BUF("aab", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("1", 1)))) => 0;
printf("btree: w%d 0x%x.%x\n",
@@ -3549,7 +3549,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF(name, 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
LFSR_DATA_BUF(&(uint8_t){'0'+(0 % 10)}, 1)))) => 0;
@@ -3564,7 +3564,7 @@ code = '''
LFSR_DATA_BUF(&(uint8_t){'0'+((i-1) % 10)}, 1)),
LFSR_ATTR(
LFSR_TAG_NAME, +1, LFSR_DATA_CAT(
LFSR_DATA_LEB128(i*DID),
LFSR_DATA_WORD(i*DID),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -3625,7 +3625,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("___", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("___", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("_", 1)))) => 0;
// set up a simulation to compare against
@@ -3666,7 +3666,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -3750,7 +3750,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +W,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF(name, 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
LFSR_DATA_BUF(&(uint8_t){'0'+(0 % 10)}, 1)))) => 0;
@@ -3766,7 +3766,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +W,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(i*DID),
LFSR_DATA_WORD(i*DID),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -3828,7 +3828,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +W,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("___", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("___", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("_", 1)))) => 0;
// set up a simulation to compare against
@@ -3885,7 +3885,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +weight2,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -3991,7 +3991,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("___", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("___", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("_", 1)))) => 0;
// set up a simulation to compare against
@@ -4043,7 +4043,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -4059,7 +4059,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +1,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -4169,7 +4169,7 @@ code = '''
lfsr_btree_commit(&lfs, &btree, 0, LFSR_ATTRS(
LFSR_ATTR(
LFSR_TAG_NAME, +W,
LFSR_DATA_CAT(LFSR_DATA_LEB128(0), LFSR_DATA_BUF("___", 3))),
LFSR_DATA_CAT(LFSR_DATA_WORD(0), LFSR_DATA_BUF("___", 3))),
LFSR_ATTR(LFSR_TAG_DATA, 0, LFSR_DATA_BUF("_", 1)))) => 0;
// set up a simulation to compare against
@@ -4233,7 +4233,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +weight,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,
@@ -4249,7 +4249,7 @@ code = '''
LFSR_ATTR(
LFSR_TAG_NAME, +weight,
LFSR_DATA_CAT(
LFSR_DATA_LEB128(0),
LFSR_DATA_WORD(0),
LFSR_DATA_BUF(name, 3))),
LFSR_ATTR(
LFSR_TAG_DATA, 0,