Make type 1 font program decryption faster. #6933
Conversation
|
decryptAscii() also probably needs a similar treatment |
brendandahl
force-pushed
the
faster-decrypt
branch
from
1139665 to
9b21584
Compare
|
Changed it to gracefully handle a bad discard number instead of asserting. |
brendandahl
force-pushed
the
faster-decrypt
branch
from
9b21584 to
a5f70c0
Compare
|
Last failure was from some debug code to find a decryptASCII use. Interesting that we only have one PDF that uses that. |
| if (discardNumber > data.length) { | ||
| return new Uint8Array(0); | ||
| } | ||
| var r = key | 0, c1 = 52845, c2 = 22719, i = 0, j = 0; |
There was a problem hiding this comment.
You already initialize i and j to 0 in the loops, so we can remove the initialization here to avoid doing it twice.
Discard the values first so we don't have to slice the array.
brendandahl
force-pushed
the
faster-decrypt
branch
from
a5f70c0 to
02331f6
Compare
| } | ||
| var r = key | 0, c1 = 52845, c2 = 22719, i, j; | ||
| for (i = 0; i < discardNumber; i++) { | ||
| r = ((data[i] + r) * c1 + c2) & ((1 << 16) - 1); |
There was a problem hiding this comment.
Given that ((1 << 16) - 1) is constant in all the loop iterations, would it make sense to cache this value in a variable instead?
There was a problem hiding this comment.
Talked with one of the spider monkey people, the value will be constant folded by the parser, so it doesn't matter.
|
@brendandahl Could you address the comment above? Also, do you want to do this for |
This one is more tricky and we only have one example, so I don't think it's worth it unless we see it show up on performance profile. |
|
/botio-linux preview |
From: Bot.io (Linux)ReceivedCommand cmd_preview from @timvandermeij received. Current queue size: 0 Live output at: http://107.21.233.14:8877/951a9d35779dbae/output.txt |
From: Bot.io (Linux)SuccessFull output at http://107.21.233.14:8877/951a9d35779dbae/output.txt Total script time: 0.81 mins Published |
|
/botio test |
From: Bot.io (Windows)ReceivedCommand cmd_test from @timvandermeij received. Current queue size: 0 Live output at: http://107.22.172.223:8877/bcfa87fea2f784e/output.txt |
From: Bot.io (Linux)ReceivedCommand cmd_test from @timvandermeij received. Current queue size: 0 Live output at: http://107.21.233.14:8877/3455e24635d82ec/output.txt |
From: Bot.io (Windows)SuccessFull output at http://107.22.172.223:8877/bcfa87fea2f784e/output.txt Total script time: 19.94 mins
|
From: Bot.io (Linux)SuccessFull output at http://107.21.233.14:8877/3455e24635d82ec/output.txt Total script time: 21.25 mins
|
Make type 1 font program decryption faster.
|
Nice work, thanks! |
Discard the values first so we don't have to slice the array.
I noticed this showed up pretty high in a profile so I checked it out. In chrome I see 31ms improvement and in firefox a 13ms improvement in the decrypt time while loading the first two pages of trace monkey.