| commit | 829519880a4ec7443d6a0a8c35fb3f3dba37f328 | [log] [tgz] |
|---|---|---|
| author | Nigel Tao <nigeltao@golang.org> | Mon Jun 01 13:31:48 2020 +1000 |
| committer | Nigel Tao <nigeltao@golang.org> | Mon Jun 01 13:59:50 2020 +1000 |
| tree | 0a9c7cbdfa3fe5f8f30a0c0fde3359a7c384fc01 | |
| parent | 4b9b9714365c0180008ba5bcfc971a6becb806c9 [diff] |
Drop unnecessary memcpy in do_bench_image_decode This doesn't affect the std/bmp or std/gif implementations per se, only the code that benches them. Prior to this commit, the source file was first decoded into a pixel buffer and then always copied from the (2-dimensional) wuffs_base__pixel_buffer to a (1-dimensional) wuffs_base_io_buffer. That copy is useful for *tests*, which compare that io_buffer's contents to a known good result (i.e. a separate I/O buffer filled with the contents of a golden file), but it isn't necessary for *benches*. Dropping that unnecessary copy reduces the total time per iteration, and therefore increases the reported speed. name old speed new speed delta wuffs_bmp_decode_40k/clang5 2.09GB/s ± 2% 2.48GB/s ± 0% +18.78% (p=0.008 n=5+5) wuffs_bmp_decode_40k/gcc7 1.73GB/s ± 1% 1.93GB/s ± 0% +11.54% (p=0.008 n=5+5) wuffs_gif_decode_1k_bw/clang5 291MB/s ± 0% 376MB/s ± 1% +29.14% (p=0.008 n=5+5) wuffs_gif_decode_1k_color_full_init/clang5 95.3MB/s ± 1% 103.2MB/s ± 1% +8.37% (p=0.008 n=5+5) wuffs_gif_decode_1k_color_part_init/clang5 117MB/s ± 0% 130MB/s ± 0% +10.68% (p=0.008 n=5+5) wuffs_gif_decode_10k_bgra/clang5 469MB/s ± 0% 489MB/s ± 1% +4.39% (p=0.008 n=5+5) wuffs_gif_decode_10k_indexed/clang5 126MB/s ± 1% 130MB/s ± 1% +3.50% (p=0.008 n=5+5) wuffs_gif_decode_20k/clang5 152MB/s ± 0% 156MB/s ± 1% +2.77% (p=0.008 n=5+5) wuffs_gif_decode_100k_artificial/clang5 318MB/s ± 0% 329MB/s ± 0% +3.53% (p=0.008 n=5+5) wuffs_gif_decode_100k_realistic/clang5 137MB/s ± 0% 139MB/s ± 0% +1.40% (p=0.008 n=5+5) wuffs_gif_decode_1000k_full_init/clang5 138MB/s ± 0% 140MB/s ± 1% +1.45% (p=0.008 n=5+5) wuffs_gif_decode_1000k_part_init/clang5 139MB/s ± 0% 140MB/s ± 1% +1.03% (p=0.008 n=5+5) wuffs_gif_decode_anim_screencap/clang5 661MB/s ± 0% 747MB/s ± 1% +13.08% (p=0.008 n=5+5) mimic_gif_decode_1k_bw/clang5 92.0MB/s ± 0% 91.9MB/s ± 1% ~ (p=0.841 n=5+5) mimic_gif_decode_1k_color/clang5 45.5MB/s ± 0% 45.7MB/s ± 0% +0.45% (p=0.032 n=5+4) mimic_gif_decode_10k_indexed/clang5 53.5MB/s ± 0% 53.7MB/s ± 0% +0.31% (p=0.032 n=5+4) mimic_gif_decode_20k/clang5 55.5MB/s ± 0% 55.1MB/s ± 2% ~ (p=0.690 n=5+5) mimic_gif_decode_100k_artificial/clang5 88.8MB/s ± 0% 89.0MB/s ± 1% ~ (p=0.151 n=5+5) mimic_gif_decode_100k_realistic/clang5 53.5MB/s ± 1% 53.7MB/s ± 0% +0.31% (p=0.024 n=5+5) mimic_gif_decode_1000k/clang5 54.6MB/s ± 0% 54.6MB/s ± 1% ~ (p=0.222 n=5+5) mimic_gif_decode_anim_screencap/clang5 106MB/s ± 0% 109MB/s ± 0% +2.59% (p=0.008 n=5+5) wuffs_gif_decode_1k_bw/gcc7 284MB/s ± 0% 356MB/s ± 0% +25.33% (p=0.016 n=4+5) wuffs_gif_decode_1k_color_full_init/gcc7 82.1MB/s ± 0% 95.3MB/s ± 0% +16.08% (p=0.008 n=5+5) wuffs_gif_decode_1k_color_part_init/gcc7 97.5MB/s ± 0% 117.4MB/s ± 0% +20.39% (p=0.008 n=5+5) wuffs_gif_decode_10k_bgra/gcc7 345MB/s ± 0% 396MB/s ± 0% +14.71% (p=0.016 n=4+5) wuffs_gif_decode_10k_indexed/gcc7 104MB/s ± 0% 121MB/s ± 0% +16.38% (p=0.008 n=5+5) wuffs_gif_decode_20k/gcc7 131MB/s ± 0% 149MB/s ± 0% +13.55% (p=0.008 n=5+5) wuffs_gif_decode_100k_artificial/gcc7 293MB/s ± 0% 321MB/s ± 1% +9.77% (p=0.008 n=5+5) wuffs_gif_decode_100k_realistic/gcc7 116MB/s ± 1% 130MB/s ± 1% +12.10% (p=0.008 n=5+5) wuffs_gif_decode_1000k_full_init/gcc7 117MB/s ± 1% 132MB/s ± 1% +12.53% (p=0.008 n=5+5) wuffs_gif_decode_1000k_part_init/gcc7 117MB/s ± 0% 132MB/s ± 1% +12.20% (p=0.008 n=5+5) wuffs_gif_decode_anim_screencap/gcc7 615MB/s ± 2% 711MB/s ± 2% +15.53% (p=0.008 n=5+5) mimic_gif_decode_1k_bw/gcc7 91.4MB/s ± 0% 91.7MB/s ± 1% ~ (p=0.095 n=5+5) mimic_gif_decode_1k_color/gcc7 46.3MB/s ± 0% 46.2MB/s ± 2% ~ (p=0.841 n=5+5) mimic_gif_decode_10k_indexed/gcc7 51.9MB/s ± 1% 51.7MB/s ± 1% ~ (p=0.310 n=5+5) mimic_gif_decode_20k/gcc7 54.2MB/s ± 0% 54.2MB/s ± 1% ~ (p=0.421 n=5+5) mimic_gif_decode_100k_artificial/gcc7 88.4MB/s ± 1% 88.7MB/s ± 1% ~ (p=0.222 n=5+5) mimic_gif_decode_100k_realistic/gcc7 53.7MB/s ± 0% 53.7MB/s ± 0% ~ (p=0.206 n=5+5) mimic_gif_decode_1000k/gcc7 53.8MB/s ± 2% 54.0MB/s ± 2% ~ (p=0.690 n=5+5) mimic_gif_decode_anim_screencap/gcc7 106MB/s ± 0% 109MB/s ± 0% +2.79% (p=0.008 n=5+5)
(Formerly known as Puffs: Parsing Untrusted File Formats Safely).
Wuffs is a memory-safe programming language (and a standard library written in that language) for wrangling untrusted file formats safely. Wrangling includes parsing, decoding and encoding. Example file formats include images, audio, video, fonts and compressed archives.
It is also fast. On many of its GIF decoding benchmarks, Wuffs measures 2x faster than “giflib” (C), 3x faster than “image/gif” (Go) and 7x faster than “gif” (Rust).
Wuffs' goal is to produce software libraries that are as safe as Go or Rust, roughly speaking, but as fast as C, and that can be used anywhere C libraries are used. This includes very large C/C++ projects, such as popular web browsers and operating systems (using that term to include desktop and mobile user interfaces, not just the kernel).
Wuffs the Library is available as transpiled C code. Other C/C++ projects can use that library without requiring the Wuffs the Language toolchain. Those projects can use Wuffs the Library like using any other third party C library. It's just not hand-written C.
However, unlike hand-written C, Wuffs the Language is safe with respect to buffer overflows, integer arithmetic overflows and null pointer dereferences. A key difference between Wuffs and other memory-safe languages is that all such checks are done at compile time, not at run time. If it compiles, it is safe, with respect to those three bug classes.
The trade-off in aiming for both safety and speed is that Wuffs programs take longer for a programmer to write, as they have to explicitly annotate their programs with proofs of safety. A statement like x += 1 unsurprisingly means to increment the variable x by 1. However, in Wuffs, such a statement is a compile time error unless the compiler can also prove that x is not the maximal value of x's type (e.g. x is not 255 if x is a base.u8), as the increment would otherwise overflow. Similarly, an integer arithmetic expression like x / y is a compile time error unless the compiler can also prove that y is not zero.
Wuffs is not a general purpose programming language. It is for writing libraries, not programs. The idea isn't to write your whole program in Wuffs, only the parts that are both performance-conscious and security-conscious. For example, while technically possible, it is unlikely that a Wuffs compiler would be worth writing entirely in Wuffs.
The /std/lzw/decode_lzw.wuffs file is a good example. The Wuffs the Language document has more information on how it differs from other languages in the C family.
For example, making this one-line edit to the LZW codec leads to a compile time error. wuffs gen fails to generate the C code, i.e. fails to compile (transpile) the Wuffs code to C code:
diff --git a/std/lzw/decode_lzw.wuffs b/std/lzw/decode_lzw.wuffs index f878c5e..f10dcee 100644 --- a/std/lzw/decode_lzw.wuffs +++ b/std/lzw/decode_lzw.wuffs @@ -98,7 +98,7 @@ pub func lzw_decoder.decode?(dst ptr buf1, src ptr buf1, src_final bool)() { in.dst.write?(x:s) if use_save_code { - this.suffixes[save_code] = c as u8 + this.suffixes[save_code] = (c + 1) as u8 this.prefixes[save_code] = prev_code as u16 }
$ wuffs gen std/gif
check: expression "(c + 1) as u8" bounds [1 ..= 256] is not within bounds [0 ..= 255] at
/home/n/go/src/github.com/google/wuffs/std/lzw/decode_lzw.wuffs:101. Facts:
n_bits < 8
c < 256
this.stack[s] == (c as u8)
use_save_code
In comparison, this two-line edit will compile (but the “does it decode GIF correctly” tests then fail):
diff --git a/std/lzw/decode_lzw.wuffs b/std/lzw/decode_lzw.wuffs index f878c5e..b43443d 100644 --- a/std/lzw/decode_lzw.wuffs +++ b/std/lzw/decode_lzw.wuffs @@ -97,8 +97,8 @@ pub func lzw_decoder.decode?(dst ptr buf1, src ptr buf1, src_final bool)() { // type checking, bounds checking and code generation for it). in.dst.write?(x:s) - if use_save_code { - this.suffixes[save_code] = c as u8 + if use_save_code and (c < 200) { + this.suffixes[save_code] = (c + 1) as u8 this.prefixes[save_code] = prev_code as u16 }
$ wuffs gen std/gif gen wrote: /home/n/go/src/github.com/google/wuffs/gen/c/gif.c gen unchanged: /home/n/go/src/github.com/google/wuffs/gen/h/gif.h $ wuffs test std/gif gen unchanged: /home/n/go/src/github.com/google/wuffs/gen/c/gif.c gen unchanged: /home/n/go/src/github.com/google/wuffs/gen/h/gif.h test: /home/n/go/src/github.com/google/wuffs/test/c/gif gif/basic.c clang PASS (8 tests run) gif/basic.c gcc PASS (8 tests run) gif/gif.c clang FAIL test_lzw_decode: bufs1_equal: wi: got 19311, want 19200. contents differ at byte 3 (in hex: 0x000003): 000000: dcdc dc00 00d9 f5f9 f6df dc5f 393a 3a3a ..........._9::: 000010: 3a3b 618e c8e4 e4e4 e5e4 e600 00e4 bbbb :;a............. 000020: eded 8f91 9191 9090 9090 9190 9192 9192 ................ 000030: 9191 9292 9191 9293 93f0 f0f0 f1f1 f2f2 ................ excerpts of got (above) versus want (below): 000000: dcdc dcdc dcd9 f5f9 f6df dc5f 393a 3a3a ..........._9::: 000010: 3a3a 618e c8e4 e4e4 e5e4 e6e4 e4e4 bbbb ::a............. 000020: eded 8f91 9191 9090 9090 9090 9191 9191 ................ 000030: 9191 9191 9191 9193 93f0 f0f0 f1f1 f2f2 ................ gif/gif.c gcc FAIL test_lzw_decode: bufs1_equal: wi: got 19311, want 19200. contents differ at byte 3 (in hex: 0x000003): 000000: dcdc dc00 00d9 f5f9 f6df dc5f 393a 3a3a ..........._9::: 000010: 3a3b 618e c8e4 e4e4 e5e4 e600 00e4 bbbb :;a............. 000020: eded 8f91 9191 9090 9090 9190 9192 9192 ................ 000030: 9191 9292 9191 9293 93f0 f0f0 f1f1 f2f2 ................ excerpts of got (above) versus want (below): 000000: dcdc dcdc dcd9 f5f9 f6df dc5f 393a 3a3a ..........._9::: 000010: 3a3a 618e c8e4 e4e4 e5e4 e6e4 e4e4 bbbb ::a............. 000020: eded 8f91 9191 9090 9090 9090 9191 9191 ................ 000030: 9191 9191 9191 9193 93f0 f0f0 f1f1 f2f2 ................ wuffs-test-c: some tests failed wuffs test: some tests failed
lang holds the Go libraries that implement Wuffs the Language: tokenizer, AST, parser, renderer, etc. The Wuffs tools are written in Go, but as mentioned above, Wuffs transpiles to C code, and Go is not necessarily involved if all you want is to use the C edition of Wuffs.lib holds other Go libraries, not specific to Wuffs the Language per se.internal holds internal implementation details, as per Go's internal packages convention.cmd holds Wuffs the Language' command line tools, also written in Go.std holds Wuffs the Library's code.release holds the releases (e.g. in their C form) of Wuffs the Library.test holds the regular tests for Wuffs the Library.fuzz holds the fuzz tests for Wuffs the Library.script holds miscellaneous utility programs.doc holds documentation.example holds example programs for Wuffs the Library.hello-wuffs-c holds an example program for Wuffs the Language.The Note directory also contains various short articles.
Version 0.2. The API and ABI aren't stabilized yet. The compiler undoubtedly has bugs. Assertion checking needs more rigor, especially around side effects and aliasing, and being sufficiently well specified to allow alternative implementations. Lots of detail needs work, but the broad brushstrokes are there.
The mailing list is at https://groups.google.com/forum/#!forum/wuffs.
The CONTRIBUTING.md file contains instructions on how to file the Contributor License Agreement before sending any pull requests (PRs). Of course, if you‘re new to the project, it’s usually best to discuss any proposals and reach consensus before sending your first PR.
Source code is auto-formatted.
Apache 2. See the LICENSE file for details.
This is not an official Google product, it is just code that happens to be owned by Google.
Updated on December 2019.