ddidderr/zstd-rs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

introduce macros STORE_OFFSET() and STORE_REPCODE()

Browse Source 
this meant to abstract the sumtype representation required
to transfert `offcode` to `ZSTD_storeSeq()`.

Unfortunately, the sumtype numeric representation is currently a leaky abstraction
that has permeated many other parts of the code,
especially within `zstd_lazy.c` and also within `zstd_opt.c` and `zstd_compress.c`.

While this PR makes a good job a transfering a large nb of call sites
to using the new macros, there are still a few sites where this transformation is more complex,
or where the numeric representation itself it used "as is".

One of the problematics area is the decision to use the numeric format of the sumtype
within the match finders of `zstd_lazy`.

This commit doesn't change the behavior, it only introduces and employes the macros,
but eventually the resulting code remains identical.

At target, if the numeric representation of the sumtype can be completely abstracted
and no other part of the code depends on it,
it will be possible to move it towards something slightly more efficient.
This commit is contained in:
Yann Collet
2021-12-23 22:03:30 -08:00
parent bec7bbb5a4
commit 1aed962216
8 changed files with 158 additions and 137 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/compress/zstd_fast.c
+10 -10
View File
@@ -180,7 +180,7 @@ _start: /* Requires: ip0 */
mLength = ip0[-1] == match0[-1];
ip0 -= mLength;
match0 -= mLength;
offcode = 0;
offcode = STORE_REPCODE_1;
mLength += 4;
goto _match;
}
@@ -267,7 +267,7 @@ _offset: /* Requires: ip0, idx */
match0 = base + idx;
rep_offset2 = rep_offset1;
rep_offset1 = (U32)(ip0-match0);
offcode = rep_offset1 + ZSTD_REP_MOVE;
offcode = STORE_OFFSET(rep_offset1);
mLength = 4;
/* Count the backwards match length. */
@@ -306,7 +306,7 @@ _match: /* Requires: ip0, match0, offcode */
{ U32 const tmpOff = rep_offset2; rep_offset2 = rep_offset1; rep_offset1 = tmpOff; } /* swap rep_offset2 <=> rep_offset1 */
hashTable[ZSTD_hashPtr(ip0, hlog, mls)] = (U32)(ip0-base);
ip0 += rLength;
ZSTD_storeSeq(seqStore, 0 /*litLen*/, anchor, iend, 0 /*offCode*/, rLength);
ZSTD_storeSeq(seqStore, 0 /*litLen*/, anchor, iend, STORE_REPCODE_1, rLength);
anchor = ip0;
continue; /* faster when present (confirmed on gcc-8) ... (?) */
} } }
@@ -439,7 +439,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, STORE_REPCODE_1, mLength);
} else if ( (matchIndex <= prefixStartIndex) ) {
size_t const dictHash = ZSTD_hashPtr(ip, dictHLog, mls);
U32 const dictMatchIndex = dictHashTable[dictHash];
@@ -459,7 +459,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
} /* catch up */
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, STORE_OFFSET(offset), mLength);
}
} else if (MEM_read32(match) != MEM_read32(ip)) {
/* it's not a match, and we're not going to check the dictionary */
@@ -474,7 +474,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
&& (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, STORE_OFFSET(offset), mLength);
}
/* match found */
@@ -499,7 +499,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2);
ZSTD_storeSeq(seqStore, 0, anchor, iend, STORE_REPCODE_1, repLength2);
hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
ip += repLength2;
anchor = ip;
@@ -598,7 +598,7 @@ static size_t ZSTD_compressBlock_fast_extDict_generic(
const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
size_t const rLength = ZSTD_count_2segments(ip+1 +4, repMatch +4, iend, repMatchEnd, prefixStart) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, rLength);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, STORE_REPCODE_1, rLength);
ip += rLength;
anchor = ip;
} else {
@@ -614,7 +614,7 @@ static size_t ZSTD_compressBlock_fast_extDict_generic(
size_t mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
offset_2 = offset_1; offset_1 = offset; /* update offset history */
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, STORE_OFFSET(offset), mLength);
ip += mLength;
anchor = ip;
} }
@@ -633,7 +633,7 @@ static size_t ZSTD_compressBlock_fast_extDict_generic(
const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
{ U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; } /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0 /*litlen*/, anchor, iend, 0 /*offcode*/, repLength2);
ZSTD_storeSeq(seqStore, 0 /*litlen*/, anchor, iend, STORE_REPCODE_1, repLength2);
hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
ip += repLength2;
anchor = ip;