GitLab

 * Register cleanup function with atexit
   instead of calling it explicitly on
   exit()
 * abort() no longer calls _cleanup:
   Flushing stdio buffers on abort is no
   longer required by POSIX.
 * dlmalloc no longer need to reset cleanup
   (see above)
 * Upstream findfp.c makebuf.c setvbuf.cexit.c
   to openbsd versions.

Bug: 14415367
Change-Id: I277058852485a9d3dbb13e5c232db5f9948d78ac

Bug: 12828904
Bug: 12875898
Change-Id: I0ec0dfe16af80bfb3153f3c3b1b3d180eff30f39

See 93e1907a

Change-Id: Icb207f4e17cfffdd6d37500656434adf1cc34427

Bug: 14468519
Change-Id: I642a5515acf5fbed46e77e176f6b5dd3b7c7bd45

This isn't declared in any header file.

Bug: 14970171
Change-Id: Ib9fce61343dfb6b6ccd7e1430e1a6e34e4e869df

This structure is huge (~18000 bytes on arm64) and can blow out
the stack very easily.
Modify the code to allocate these structures instead of leaving them
on the stack.

Bug: 14468519
Change-Id: I774f71235d896d32a14ab1af06f95ca9ef819f52

This hasn't built in over one release cycle and no one even noticed.
art does this the right way and other projects should do the same.

Change-Id: I7d1fb84c4080e008f329ee73e209ce85a36e6d55

(cherry picked from commit 58b1f3f6a30a660ad81637c2b50382c3d279243b)

Change-Id: I5d09be413cf720fbed905f96313b007997ada76c

Change-Id: I79261de70d225236d0eadff288220258d697437f

That's what the Google style guide recommends, and we're starting
to get a mix.

Change-Id: Ib0c53a890bb5deed5c679e887541a715faea91fc

Glibc calls the tag pointer in user_fpregs_struct ftw instead of twd.

Change-Id: I4a8dab777fa889de0b9927035cd18557ae6f3263

Stupidly I found this bug by accident when writing the existing
tests, but I didn't think any real code would hit it. It turns
out that libcore always uses an INET6_ADDRSTRLEN-sized buffer
even when working with AF_INET addresses.

Change-Id: Ieffc8e4bbe9b66b49b033e3e7101c896e097e6f8

Change-Id: I833b51318f0bf7749073859b7ffacc668c27558e

Use the upstream OpenBSD implementations of these functions.

Also ensure we have symbols for htonl, htons, ntohl, and ntohs.
gtest doesn't like us using the macro versions in ASSERT_EQ.

Bug: 14840760
Change-Id: I68720e9aca14838df457d2bb27b999d5818ac2b5

Change-Id: I3f874326d96f41249f5c9b1b3c2885f858589029

Make sure __netdClientDispatch is defined in the same set of libraries that
refer to it (e.g.: with connect.cpp).

Change-Id: I86d7bf2df5bde09f75a35b204eac0e1361747e22

Change-Id: I33293d8bc62cbb22e23a704c4242e7e9d3fce7c5

The library exists outside bionic. It is dynamically loaded, to replace selected
standard socket syscalls with versions that talk to netd.

Change connect() to use the library if available.

(cherry picked from commit 3a6b627a14df8111b03e452f2df4b5f4938e0e49)

Change-Id: Ib6198e19dbc306521a26fcecfdf6e8424d163fc9

From the release notes:

  Changes affecting near-future time stamps

    Egypt observes DST starting 2014-05-15 at 24:00.
    (Thanks to Ahmad El-Dardiry and Gunther Vermier.)
    Details have not been announced, except that DST will not be
    observed during Ramadan.  Guess that DST will stop during the same
    Ramadan dates as Morocco, and that Egypt's future spring and fall
    transitions will be the same as 2010 when it last observed DST,
    namely April's last Friday at 00:00 to September's last Thursday at
    23:00 standard time.  Also, guess that Ramadan transitions will be
    at 00:00 standard time.

Change-Id: I6a20cae02a314871acbd52cb90fcbebd37625810

This helps build 'external/flo' out of the box.

Bug: 14841211
Change-Id: I30dde77239cceaf1f5743163744eb3604d27a266

Also ensure that none of our home-grown code uses more than 2KiB per frame.

Change-Id: I8987a17d72f4b7f082bb7fa25e137c8433664c14

Gcc defines only __LP64__ so testing __LP32__ will always be false.

Bug: 14881256
Change-Id: Ic63c1d562be09c3eb4b9e17c9e5ef2d9e404db80

On LP64 this brings us on par with the other libcs where wctype_t is 8
bytes.

Bug: 12875898
Change-Id: Ice4f538ccf0634ef6667a8d90d0f7f09cec9e1b0

This increases bionic source compatibility with other libcs where
"wctype_t foo = 0;" is valid without -fpermissive.

Bug: 14646243
Change-Id: Ia9bd0785bc42c7b46e2bb6c3d9b9a9d3f769d983

Bug: 14865741
Change-Id: I1398f7b3f64e3c94f2714cede5b61205dfa78a24

Add following functions:
bcopy, bzero, memcpy, memmove, memset, stpcpy, stpncpy, strcat, strcpy,
strlen, strncat, strncpy, memcmp, strcmp, strncmp.
Set all these functions as the default ones.

Change-Id: Ic66b250ad8c349a43d25e2d4dea075604f6df6ac
Signed-off-by: Varvara Rainchik <varvara.rainchik@intel.com>

Bug: 13077905
Change-Id: I5abdc7cc3c27c109b7900c94b112f18a95c35763

Change-Id: I12e9d6716c42ccbccc9a186441aca0736bb22d05

Add following functions:
bcopy, memcpy, memmove, memset, bzero, memcmp, wmemcmp, strlen,
strcpy, strncpy, stpcpy, stpncpy.
Create new directories inside arch-x86 to specify architecture: atom,
silvermont and generic (non atom or silvermont architectures are treated like generic).
Due to introducing optimized versions of stpcpy and stpncpy,
c-implementations of these functions are moved from
common for architectures makefile to arm and mips specific makefiles.

Change-Id: I990f8061c3e9bca1f154119303da9e781c5d086e
Signed-off-by: Varvara Rainchik <varvara.rainchik@intel.com>

The kernel struct has some spare space at the end too, and some extra
fields, so having a bit of flexibility in statvfs might be worthwhile.

Bug: 14681331
Bug: 12875898
Change-Id: I5b502c5dd9d2e3bb8f34804f83c02669cefce01e

I've been meaning to do this for a very long time...

Change-Id: Ia8c16eee7c026c3c9505399948485fb778fb0152

Also let clone(2) set the TLS for x86.

Also ensure we initialize the TLS before we clone(2) for all architectures.

Change-Id: Ie5fa4466e1c9ee116a281dfedef574c5ba60c0b5

Add pg_gecos ("real name") field to struct passwd.

Bug: 14679430
Bug: 12875898
Change-Id: I833ec2c46f6a6259b7232a8e3f7942968ef0b50d

Also ensure that arm64/x86-64/x86 assembler uses local labels.
(There are are so many non-local labels in arm that fixing them
seems out of scope.)

Also synchronize the __bionic_clone.S comments.

Change-Id: I03b4f84780d996b54d6637a074638196bbb01cd4

Included is a new target generic-neon that will use neon instructions
on 64 bit platforms where appropriate.

Change-Id: Iaf71b768780aa9240a24539cd87666ca9298e4c6

clone(2) is the public symbol.

Also switch a test from __bionic_clone to clone; testing public API
means the test now works on glibc too.

Change-Id: If59def26a00c3afadb8a6cf9442094c35a59ffde

Change-Id: I66517e30c568d7f36f6bf6b1e4507ecf14f01226

Bug: 14159417
Change-Id: I7aa986fb8e11660035221f9bb4fe0bc40ac36c58

This matches what frameworks/base does with Build.VERSION and means that
bionic's version number will always sort >= than any released version.
This should prevent confusion in code that builds both against bionic
and the NDK.

(Note that <sys/cdefs.h> drags this in, so it's always in the namespace.)

Bug: 14613709
Change-Id: I91fb745920e848a6b20f2f5797c0a7d6cde6c032

This more general interface lets liblog give us any fatal log message,
regardless of source. This means we can remove the special case for
LOG_ALWAYS_FATAL with a simpler scheme that automatically works for
the VM too.

Change-Id: Ia6dbf7c3dbabf223081bd5159294835d954bb067