1Correctness Testing 2=================== 3 4Skia correctness testing is primarily served by a tool named DM. 5This is a quickstart to building and running DM. 6 7<!--?prettify lang=sh?--> 8 9 python tools/git-sync-deps 10 bin/gn gen out/Debug 11 ninja -C out/Debug dm 12 out/Debug/dm -v -w dm_output 13 14When you run this, you may notice your CPU peg to 100% for a while, then taper 15off to 1 or 2 active cores as the run finishes. This is intentional. DM is 16very multithreaded, but some of the work, particularly GPU-backed work, is 17still forced to run on a single thread. You can use `--threads N` to limit DM to 18N threads if you like. This can sometimes be helpful on machines that have 19relatively more CPU available than RAM. 20 21As DM runs, you ought to see a giant spew of output that looks something like this. 22~~~ 23Skipping nonrendering: Don't understand 'nonrendering'. 24Skipping angle: Don't understand 'angle'. 25Skipping nvprmsaa4: Could not create a surface. 26492 srcs * 3 sinks + 382 tests == 1858 tasks 27 28( 25MB 1857) 1.36ms 8888 image mandrill_132x132_12x12.astc-5-subsets 29( 25MB 1856) 1.41ms 8888 image mandrill_132x132_6x6.astc-5-subsets 30( 25MB 1855) 1.35ms 8888 image mandrill_132x130_6x5.astc-5-subsets 31( 25MB 1854) 1.41ms 8888 image mandrill_132x130_12x10.astc-5-subsets 32( 25MB 1853) 151µs 8888 image mandrill_130x132_10x6.astc-5-subsets 33( 25MB 1852) 154µs 8888 image mandrill_130x130_5x5.astc-5-subsets 34 ... 35( 748MB 5) 9.43ms unit test GLInterfaceValidation 36( 748MB 4) 30.3ms unit test HalfFloatTextureTest 37( 748MB 3) 31.2ms unit test FloatingPointTextureTest 38( 748MB 2) 32.9ms unit test DeferredCanvas_GPU 39( 748MB 1) 49.4ms unit test ClipCache 40( 748MB 0) 37.2ms unit test Blur 41~~~ 42Do not panic. 43 44As you become more familiar with DM, this spew may be a bit annoying. If you 45remove -v from the command line, DM will spin its progress on a single line 46rather than print a new line for each status update. 47 48Don't worry about the "Skipping something: Here's why." lines at startup. DM 49supports many test configurations, which are not all appropriate for all 50machines. These lines are a sort of FYI, mostly in case DM can't run some 51configuration you might be expecting it to run. 52 53Don't worry about the "skps: Couldn't read skps." messages either, you won't 54have those by default and can do without them. If you wish to test with them 55too, you can download them separately. 56 57The next line is an overview of the work DM is about to do. 58~~~ 59492 srcs * 3 sinks + 382 tests == 1858 tasks 60~~~ 61 62DM has found 382 unit tests (code linked in from tests/), and 492 other drawing 63sources. These drawing sources may be GM integration tests (code linked in 64from gm/), image files (from `--images`, which defaults to "resources") or .skp 65files (from `--skps`, which defaults to "skps"). You can control the types of 66sources DM will use with `--src` (default, "tests gm image skp"). 67 68DM has found 3 usable ways to draw those 492 sources. This is controlled by 69`--config`. The defaults are operating system dependent. On Linux they are "8888 gl nonrendering". 70DM has skipped nonrendering leaving two usable configs: 718888 and gl. These two name different ways to draw using Skia: 72 73 - 8888: draw using the software backend into a 32-bit RGBA bitmap 74 - gl: draw using the OpenGL backend (Ganesh) into a 32-bit RGBA bitmap 75 76Sometimes DM calls these configs, sometimes sinks. Sorry. There are many 77possible configs but generally we pay most attention to 8888 and gl. 78 79DM always tries to draw all sources into all sinks, which is why we multiply 80492 by 3. The unit tests don't really fit into this source-sink model, so they 81stand alone. A couple thousand tasks is pretty normal. Let's look at the 82status line for one of those tasks. 83~~~ 84( 25MB 1857) 1.36ms 8888 image mandrill_132x132_12x12.astc-5-subsets 85~~~ 86 87This status line tells us several things. 88 89First, it tells us that at the time we wrote the status line, the maximum 90amount of memory DM had ever used was 25MB. Note this is a high water mark, 91not the current memory usage. This is mostly useful for us to track on our 92buildbots, some of which run perilously close to the system memory limit. 93 94Next, the status line tells us that there are 1857 unfinished tasks, either 95currently running or waiting to run. We generally run one task per hardware 96thread available, so on a typical laptop there are probably 4 or 8 running at 97once. Sometimes the counts appear to show up out of order, particularly at DM 98startup; it's harmless, and doesn't affect the correctness of the run. 99 100Next, we see this task took 1.36 milliseconds to run. Generally, the precision 101of this timer is around 1 microsecond. The time is purely there for 102informational purposes, to make it easier for us to find slow tests. 103 104Finally we see the configuration and name of the test we ran. We drew the test 105"mandrill_132x132_12x12.astc-5-subsets", which is an "image" source, into an 106"8888" sink. 107 108When DM finishes running, you should find a directory with file named dm.json, 109and some nested directories filled with lots of images. 110~~~ 111$ ls dm_output 1128888 dm.json gl 113 114$ find dm_output -name '*.png' 115dm_output/8888/gm/3x3bitmaprect.png 116dm_output/8888/gm/aaclip.png 117dm_output/8888/gm/aarectmodes.png 118dm_output/8888/gm/alphagradients.png 119dm_output/8888/gm/arcofzorro.png 120dm_output/8888/gm/arithmode.png 121dm_output/8888/gm/astcbitmap.png 122dm_output/8888/gm/bezier_conic_effects.png 123dm_output/8888/gm/bezier_cubic_effects.png 124dm_output/8888/gm/bezier_quad_effects.png 125 ... 126~~~ 127 128The directories are nested first by sink type (`--config`), then by source type (`--src`). 129The image from the task we just looked at, "8888 image mandrill_132x132_12x12.astc-5-subsets", 130can be found at dm_output/8888/image/mandrill_132x132_12x12.astc-5-subsets.png. 131 132dm.json is used by our automated testing system, so you can ignore it if you 133like. It contains a listing of each test run and a checksum of the image 134generated for that run. 135 136### Detail <a name="digests"></a> 137Boring technical detail: The checksum is not a checksum of the 138.png file, but rather a checksum of the raw pixels used to create that .png. 139That means it is possible for two different configurations to produce 140the same exact .png, but have their checksums differ. 141 142Unit tests don't generally output anything but a status update when they pass. 143If a test fails, DM will print out its assertion failures, both at the time 144they happen and then again all together after everything is done running. 145These failures are also included in the dm.json file. 146 147DM has a simple facility to compare against the results of a previous run: 148 149<!--?prettify lang=sh?--> 150 151 ninja -C out/Debug dm 152 out/Debug/dm -w good 153 154 # do some work 155 156 ninja -C out/Debug dm 157 out/Debug/dm -r good -w bad 158 159When using `-r`, DM will display a failure for any test that didn't produce the 160same image as the `good` run. 161 162For anything fancier, I suggest using skdiff: 163 164<!--?prettify lang=sh?--> 165 166 ninja -C out/Debug dm 167 out/Debug/dm -w good 168 169 # do some work 170 171 ninja -C out/Debug dm 172 out/Debug/dm -w bad 173 174 ninja -C out/Debug skdiff 175 mkdir diff 176 out/Debug/skdiff good bad diff 177 178 # open diff/index.html in your web browser 179 180That's the basics of DM. DM supports many other modes and flags. Here are a 181few examples you might find handy. 182 183<!--?prettify lang=sh?--> 184 185 out/Debug/dm --help # Print all flags, their defaults, and a brief explanation of each. 186 out/Debug/dm --src tests # Run only unit tests. 187 out/Debug/dm --nocpu # Test only GPU-backed work. 188 out/Debug/dm --nogpu # Test only CPU-backed work. 189 out/Debug/dm --match blur # Run only work with "blur" in its name. 190 out/Debug/dm --dryRun # Don't really do anything, just print out what we'd do. 191 192
