Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 1 | Working on bionic |
| 2 | ================= |
| 3 | |
| 4 | What are the big pieces of bionic? |
| 5 | ---------------------------------- |
| 6 | |
Dan Albert | 472cce5 | 2014-10-10 17:14:37 -0700 | [diff] [blame] | 7 | #### libc/ --- libc.so, libc.a |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 8 | |
Dan Albert | 472cce5 | 2014-10-10 17:14:37 -0700 | [diff] [blame] | 9 | The C library. Stuff like `fopen(3)` and `kill(2)`. |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 10 | |
Dan Albert | 472cce5 | 2014-10-10 17:14:37 -0700 | [diff] [blame] | 11 | #### libm/ --- libm.so, libm.a |
| 12 | |
| 13 | The math library. Traditionally Unix systems kept stuff like `sin(3)` and |
| 14 | `cos(3)` in a separate library to save space in the days before shared |
| 15 | libraries. |
| 16 | |
| 17 | #### libdl/ --- libdl.so |
| 18 | |
| 19 | The dynamic linker interface library. This is actually just a bunch of stubs |
| 20 | that the dynamic linker replaces with pointers to its own implementation at |
| 21 | runtime. This is where stuff like `dlopen(3)` lives. |
| 22 | |
| 23 | #### libstdc++/ --- libstdc++.so |
| 24 | |
| 25 | The C++ ABI support functions. The C++ compiler doesn't know how to implement |
| 26 | thread-safe static initialization and the like, so it just calls functions that |
| 27 | are supplied by the system. Stuff like `__cxa_guard_acquire` and |
| 28 | `__cxa_pure_virtual` live here. |
| 29 | |
| 30 | #### linker/ --- /system/bin/linker and /system/bin/linker64 |
| 31 | |
| 32 | The dynamic linker. When you run a dynamically-linked executable, its ELF file |
| 33 | has a `DT_INTERP` entry that says "use the following program to start me". On |
| 34 | Android, that's either `linker` or `linker64` (depending on whether it's a |
| 35 | 32-bit or 64-bit executable). It's responsible for loading the ELF executable |
| 36 | into memory and resolving references to symbols (so that when your code tries to |
| 37 | jump to `fopen(3)`, say, it lands in the right place). |
| 38 | |
| 39 | #### tests/ --- unit tests |
| 40 | |
| 41 | The `tests/` directory contains unit tests. Roughly arranged as one file per |
| 42 | publicly-exported header file. |
| 43 | |
| 44 | #### benchmarks/ --- benchmarks |
| 45 | |
| 46 | The `benchmarks/` directory contains benchmarks. |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 47 | |
| 48 | |
| 49 | What's in libc/? |
| 50 | ---------------- |
| 51 | |
Dan Albert | 472cce5 | 2014-10-10 17:14:37 -0700 | [diff] [blame] | 52 | <pre> |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 53 | libc/ |
| 54 | arch-arm/ |
| 55 | arch-arm64/ |
| 56 | arch-common/ |
| 57 | arch-mips/ |
| 58 | arch-mips64/ |
| 59 | arch-x86/ |
| 60 | arch-x86_64/ |
| 61 | # Each architecture has its own subdirectory for stuff that isn't shared |
| 62 | # because it's architecture-specific. There will be a .mk file in here that |
| 63 | # drags in all the architecture-specific files. |
| 64 | bionic/ |
| 65 | # Every architecture needs a handful of machine-specific assembler files. |
| 66 | # They live here. |
| 67 | include/ |
| 68 | machine/ |
| 69 | # The majority of header files are actually in libc/include/, but many |
| 70 | # of them pull in a <machine/something.h> for things like limits, |
| 71 | # endianness, and how floating point numbers are represented. Those |
| 72 | # headers live here. |
| 73 | string/ |
| 74 | # Most architectures have a handful of optional assembler files |
| 75 | # implementing optimized versions of various routines. The <string.h> |
| 76 | # functions are particular favorites. |
| 77 | syscalls/ |
| 78 | # The syscalls directories contain script-generated assembler files. |
| 79 | # See 'Adding system calls' later. |
| 80 | |
| 81 | include/ |
| 82 | # The public header files on everyone's include path. These are a mixture of |
| 83 | # files written by us and files taken from BSD. |
| 84 | |
| 85 | kernel/ |
| 86 | # The kernel uapi header files. These are scrubbed copies of the originals |
| 87 | # in external/kernel-headers/. These files must not be edited directly. The |
| 88 | # generate_uapi_headers.sh script should be used to go from a kernel tree to |
| 89 | # external/kernel-headers/ --- this takes care of the architecture-specific |
| 90 | # details. The update_all.py script should be used to regenerate bionic's |
| 91 | # scrubbed headers from external/kernel-headers/. |
| 92 | |
| 93 | private/ |
| 94 | # These are private header files meant for use within bionic itself. |
| 95 | |
Calin Juravle | bd33537 | 2014-02-28 16:31:04 +0000 | [diff] [blame] | 96 | dns/ |
| 97 | # Contains the DNS resolver (originates from NetBSD code). |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 98 | |
| 99 | upstream-dlmalloc/ |
| 100 | upstream-freebsd/ |
| 101 | upstream-netbsd/ |
| 102 | upstream-openbsd/ |
| 103 | # These directories contain unmolested upstream source. Any time we can |
| 104 | # just use a BSD implementation of something unmodified, we should. |
Elliott Hughes | d39f3f2 | 2014-04-21 17:13:46 -0700 | [diff] [blame] | 105 | # The structure under these directories mimics the upstream tree, |
| 106 | # but there's also... |
| 107 | android/ |
| 108 | include/ |
| 109 | # This is where we keep the hacks necessary to build BSD source |
| 110 | # in our world. The *-compat.h files are automatically included |
| 111 | # using -include, but we also provide equivalents for missing |
| 112 | # header/source files needed by the BSD implementation. |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 113 | |
| 114 | bionic/ |
| 115 | # This is the biggest mess. The C++ files are files we own, typically |
| 116 | # because the Linux kernel interface is sufficiently different that we |
| 117 | # can't use any of the BSD implementations. The C files are usually |
| 118 | # legacy mess that needs to be sorted out, either by replacing it with |
| 119 | # current upstream source in one of the upstream directories or by |
| 120 | # switching the file to C++ and cleaning it up. |
| 121 | |
Elliott Hughes | 3ad8ecb | 2014-07-21 16:35:24 -0700 | [diff] [blame] | 122 | stdio/ |
| 123 | # These are legacy files of dubious provenance. We're working to clean |
| 124 | # this mess up, and this directory should disappear. |
| 125 | |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 126 | tools/ |
| 127 | # Various tools used to maintain bionic. |
| 128 | |
| 129 | tzcode/ |
| 130 | # A modified superset of the IANA tzcode. Most of the modifications relate |
| 131 | # to Android's use of a single file (with corresponding index) to contain |
| 132 | # time zone data. |
| 133 | zoneinfo/ |
| 134 | # Android-format time zone data. |
| 135 | # See 'Updating tzdata' later. |
Dan Albert | 472cce5 | 2014-10-10 17:14:37 -0700 | [diff] [blame] | 136 | </pre> |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 137 | |
| 138 | |
| 139 | Adding system calls |
| 140 | ------------------- |
| 141 | |
| 142 | Adding a system call usually involves: |
| 143 | |
| 144 | 1. Add entries to SYSCALLS.TXT. |
| 145 | See SYSCALLS.TXT itself for documentation on the format. |
| 146 | 2. Run the gensyscalls.py script. |
| 147 | 3. Add constants (and perhaps types) to the appropriate header file. |
| 148 | Note that you should check to see whether the constants are already in |
| 149 | kernel uapi header files, in which case you just need to make sure that |
Elliott Hughes | 247904a | 2014-02-21 16:09:27 -0800 | [diff] [blame] | 150 | the appropriate POSIX header file in libc/include/ includes the |
Elliott Hughes | 560cee6 | 2014-02-18 22:08:56 -0800 | [diff] [blame] | 151 | relevant file or files. |
| 152 | 4. Add function declarations to the appropriate header file. |
| 153 | 5. Add at least basic tests. Even a test that deliberately supplies |
| 154 | an invalid argument helps check that we're generating the right symbol |
| 155 | and have the right declaration in the header file. (And strace(1) can |
| 156 | confirm that the correct system call is being made.) |
| 157 | |
| 158 | |
| 159 | Updating kernel header files |
| 160 | ---------------------------- |
| 161 | |
| 162 | As mentioned above, this is currently a two-step process: |
| 163 | |
| 164 | 1. Use generate_uapi_headers.sh to go from a Linux source tree to appropriate |
| 165 | contents for external/kernel-headers/. |
| 166 | 2. Run update_all.py to scrub those headers and import them into bionic. |
| 167 | |
| 168 | |
| 169 | Updating tzdata |
| 170 | --------------- |
| 171 | |
| 172 | This is fully automated: |
| 173 | |
| 174 | 1. Run update-tzdata.py. |
| 175 | |
Dan Albert | efee1ce | 2014-10-09 22:57:49 -0700 | [diff] [blame] | 176 | |
Dan Albert | e66d57f | 2014-11-12 17:08:38 -0800 | [diff] [blame] | 177 | Verifying changes |
| 178 | ----------------- |
| 179 | |
| 180 | If you make a change that is likely to have a wide effect on the tree (such as a |
| 181 | libc header change), you should run `make checkbuild`. A regular `make` will |
| 182 | _not_ build the entire tree; just the minimum number of projects that are |
| 183 | required for the device. Tests, additional developer tools, and various other |
| 184 | modules will not be built. Note that `make checkbuild` will not be complete |
| 185 | either, as `make tests` covers a few additional modules, but generally speaking |
| 186 | `make checkbuild` is enough. |
| 187 | |
| 188 | |
Dan Albert | efee1ce | 2014-10-09 22:57:49 -0700 | [diff] [blame] | 189 | Running the tests |
| 190 | ----------------- |
| 191 | |
| 192 | The tests are all built from the tests/ directory. |
| 193 | |
| 194 | ### Device tests |
| 195 | |
| 196 | $ mma |
| 197 | $ adb sync |
| 198 | $ adb shell /data/nativetest/bionic-unit-tests/bionic-unit-tests32 |
| 199 | $ adb shell \ |
| 200 | /data/nativetest/bionic-unit-tests-static/bionic-unit-tests-static32 |
| 201 | # Only for 64-bit targets |
| 202 | $ adb shell /data/nativetest/bionic-unit-tests/bionic-unit-tests64 |
| 203 | $ adb shell \ |
| 204 | /data/nativetest/bionic-unit-tests-static/bionic-unit-tests-static64 |
| 205 | |
| 206 | ### Host tests |
| 207 | |
| 208 | The host tests require that you have `lunch`ed either an x86 or x86_64 target. |
| 209 | |
| 210 | $ mma |
| 211 | # 64-bit tests for 64-bit targets, 32-bit otherwise. |
| 212 | $ mm bionic-unit-tests-run-on-host |
| 213 | # Only exists for 64-bit targets. |
| 214 | $ mm bionic-unit-tests-run-on-host32 |
| 215 | |
| 216 | ### Against glibc |
| 217 | |
| 218 | As a way to check that our tests do in fact test the correct behavior (and not |
| 219 | just the behavior we think is correct), it is possible to run the tests against |
| 220 | the host's glibc. |
| 221 | |
| 222 | $ mma |
| 223 | $ bionic-unit-tests-glibc32 # already in your path |
| 224 | $ bionic-unit-tests-glibc64 |
| 225 | |
| 226 | |
| 227 | Gathering test coverage |
| 228 | ----------------------- |
| 229 | |
| 230 | For either host or target coverage, you must first: |
| 231 | |
| 232 | * `$ export NATIVE_COVERAGE=true` |
| 233 | * Note that the build system is ignorant to this flag being toggled, i.e. if |
| 234 | you change this flag, you will have to manually rebuild bionic. |
| 235 | * Set `bionic_coverage=true` in `libc/Android.mk` and `libm/Android.mk`. |
| 236 | |
| 237 | ### Coverage from device tests |
| 238 | |
| 239 | $ mma |
| 240 | $ adb sync |
| 241 | $ adb shell \ |
| 242 | GCOV_PREFIX=/data/local/tmp/gcov \ |
| 243 | GCOV_PREFIX_STRIP=`echo $ANDROID_BUILD_TOP | grep -o / | wc -l` \ |
| 244 | /data/nativetest/bionic-unit-tests/bionic-unit-tests32 |
| 245 | $ acov |
| 246 | |
| 247 | `acov` will pull all coverage information from the device, push it to the right |
| 248 | directories, run `lcov`, and open the coverage report in your browser. |
| 249 | |
| 250 | ### Coverage from host tests |
| 251 | |
| 252 | First, build and run the host tests as usual (see above). |
| 253 | |
| 254 | $ croot |
| 255 | $ lcov -c -d $ANDROID_PRODUCT_OUT -o coverage.info |
| 256 | $ genhtml -o covreport coverage.info # or lcov --list coverage.info |
| 257 | |
| 258 | The coverage report is now available at `covreport/index.html`. |
Elliott Hughes | 0b1de06 | 2015-01-09 12:21:24 -0800 | [diff] [blame] | 259 | |
| 260 | |
| 261 | LP32 ABI bugs |
| 262 | ------------- |
| 263 | |
| 264 | This probably belongs in the NDK documentation rather than here, but these |
| 265 | are the known ABI bugs in LP32: |
| 266 | |
| 267 | * time_t is 32-bit. http://b/5819737 |
| 268 | |
| 269 | * off_t is 32-bit. There is off64_t, but no _FILE_OFFSET_BITS support. |
| 270 | Many of the off64_t functions are missing in older releases, and |
| 271 | stdio uses 32-bit offsets, so there's no way to fully implement |
| 272 | _FILE_OFFSET_BITS. |
| 273 | |
| 274 | * sigset_t is too small on ARM and x86 (but correct on MIPS), so support |
| 275 | for real-time signals is broken. http://b/5828899 |