The usecase here is system/core/liblog, which wants to turn off
sanitizer only for the target module, since the target sanitizers depend
on liblog.
Change-Id: I99141574a1c4c4847aa8a72bdcc88901fd9b01cd
Store product variables in pointers so that we can only apply the
properties if the product variable was set to a value. Also only apply
bool properties if they are true, adn rearrange the code to do the
cheapest checks first.
Remove device_uses_logd, it doesn't exist any more.
Change-Id: Icf42408f57bd611746f8d985bfceb50c7f95ea59
Any reads after the deps are saved won't show up as dependencies later.
So panic if new environment variables are read after saving deps.
Change-Id: Ia51deaf750804d3b99e69c001939a104c2d8c9f2
Instead of using this hardcoded list, use the blueprint plugin
infrastructure to find them at build time and load them.
Bug: 23567252
Change-Id: Iae87e8c9bf020f4beb42c418b9a6bf8b0fe8b601
Refactor the soong.config loading code to support reading in
product variables from soong.variables.
Change-Id: I389e6bb5c501b53167267d5f5d0d25557811cf72
Allow modules to vary their properties based on product variables.
For now, DEVICE_USES_LOGD, DEVICE_USES_JEMALLOC, and DEVICE_USES_DLMALLOC,
and BOARD_MALLOC_ALIGNMENT are supported.
Product variables can provide a value (only bool and int supported for
now), and if any of the product variable properties contains a "%d"
then Sprintf will be called with the property value as the format
and the product variable value convert to an int as the only argument.
For example:
product_variables: {
dlmalloc_alignment: {
cflags: ["-DMALLOC_ALIGNMENT=%d"],
},
},
will cause -DMALLOC_ALIGNMENT=16 to be added to any top level
properties called "cflags".
Change-Id: I74882a6ab4914d3e222f8d06cfac371b7b829ae5
ARCH_X86_HAVE_SSE* should not have second_arch set, they always
apply to x86 regardless of TARGET_ARCH vs. TARGET_2ND_ARCH.
Change-Id: I21fa4c88e9c966a7181f3b17c95c06080b678556
The source path was being appended to the module out directory
to create the file list file, which was resulting in .. in the
source path moving the file list file up the directory tree.
Use SrcDirRelPath to convert the globbed resource directories
to be relatiave to $srcDir before appending them.
Also do the same fix to generated aidl, logtags, yacc, and lex
files.
Change-Id: I2e636bd30abf03bc1d80a897951a9812cc3e09ef
* commit 'ff93bddda3268edc4159aa2eea9d8d814727732f': (173 commits)
Fix glob filename overlap
Fix arm cflags
Add support for x86_sse3 and x86_sse4 arch sections
use init functions to register module types, etc.
Use same libgcc as Make for arm
Fix libgcc/libatomic link order
Enable host multilib
Add cpu-variant properties
Refactor standard and rewrite properties to reduce duplication
Add version_script property
androidbp: Optimize cc_library builds, share static lib
androidbp: Implement static/shared property copying
Add exclude_* and remove arch_subtract / "-file"
Use += for both assignments to LOCAL_C_INCLUDES
androidbp: simplify translation by evaluating all expressions in Blueprint
Ignore some module types during translation
Allow manually specifying translations for modules
androidbp: make error handling stricter
Remove comments from translated Android.mk files
Support the 'host' conditional in androidbp
...
If resources and java files were compiled from the same directory,
ctx.Glob could try to create a glob file that had the same name
as the directory containing another glob file. Namespace each
call to ctx.Glob so they never conflict.
Change-Id: I4db73af568a2ff2e708e9db64798073b1ed2ff61
-mthumb-interwork was missing from the beginning.
The Clang cflags copy was only a shallow copy, so our clang-specific
fixups were affecting the GCC cflags as well.
The cortex-a15 replacement was removed from build/core with this change:
https://android-review.googlesource.com/#/c/143851
Change-Id: Ib6300e064f6ec4f6a7174d5e2aeacad1c359f7ea
