Age | Commit message (Collapse) | Author |
|
Make it build if _GNU_SOURCE isn't defined by the compiler itself
Signed-off-by: Bernhard Rosenkraenzer <Bernhard.Rosenkranzer@linaro.org>
|
|
Since the key cord for screen shot is now strictly enforced, we no
longer get a free screen shot with bugreport. This change to bugreport
script will capture a screen shot first, then proceed to dumpstate.
SendBug is also updated accordingly to take 2 parameters: one for
bugreport path, the other for screen shot path.
Change-Id: I20fcea2f3a207b0dcd5341c8646c694dac12ec54
|
|
Add a small test to verify that socket tagging is not hopelessly
broken.
Test for
- invalid: arg count, command, tags.
- double tagging
- acct_tag changing
- optional uid
- 64bit unsigned vs signed
- invalid sock_fd
- multiple entries
- untagging
- uid/gid checking + max tags.
It tries its best to allow concurrent invocations, but it just
can not be sure that it always detects the in-kernel addresses of
the sockets it has tagged.
Change-Id: I4e372fcddbd0b1cd6ba5d8be4692e0f31ec71a99
|
|
Only works for email bugreport for now. Might add BugReported support later - requiers either
changes to BugReporter tool or zipping the file with is not great.
Change-Id: I1941a8436f6b5ba31b04334a89cad075f3bc836e
|
|
Queries PackageManager for a predefined intent, if it's resolvable,
hand the bugreport over to BugReporter. Otherwise same as before:
compose an intent to send out in mail with attachment
Change-Id: Ib7a5b49551ae9285ed46464277168487520d798b
|
|
Change-Id: I1247ba30e6e2e424a4cce12063a3c6af5db035e9
|
|
First script calls dumpsys to generate the bugreport, second
script launches a cmd line tool that queries account manager
for possible accounts to set as "to:" field and launches an
intent to mail the bugreport out
This is done in 2 steps because it's more convenient to look
up account for emailing with a framework context, but to make
sure bugreport is captured, using dumpsys from shell script
is more reliable in case Dalvik VM crashes.
Change-Id: Ibb54ac793bbd59f016db9fe98118155168c9371e
|
|
Now print a sum of all pss and uss at then end of the output, and
then print information from /proc/meminfo to help put those in
context.
Change-Id: I323d219820830e9435bb4dcfd78cbb9a3d6c129c
|
|
|
|
Fixed a bug where maps were being coalesced even when they were
not necessarily coalesced.
Enhanced the BSS segment detection heuristic so that it takes
the map start / end addresses into account and displays the
inferred map type directly.
Automatically sort all maps by name or address as appropriate.
Fixed a bug parsing maps with paths that contain whitespace.
Enhanced the -a argument to show the same columns as the normal
mode but just prefixed with the virtual memory map information.
Change-Id: Ice78afb0c5b597683cb1a1ba65e3c10f10258f7c
|
|
|
|
If it's not a multiple of the block size, then errors are generated
when making sparse images. This doesn't change the size of the generated
filesystem, because partial blocks are not included in the filesystem.
Change-Id: Ifc00ef7961f651ce22c6400759eadce2775d7811
|
|
|
|
Instead of handling maps with the name "[vectors]" specially,
silently ignore EOF when reading from /proc/<pid>/pagemap, which
occurs any time a a mapping is outside of the userspace range.
Change-Id: I674ade1eab6fd7732c6d9e120d0750cca5415b25
|
|
* changes:
Ignore the vector page in /proc/<pid>/maps
Detect an error case and prevent printing an uninitialized variable
|
|
In kernel 2.6.37, the vector page was added to /proc/<pid>/maps,
but because it is located above TASK_SIZE (usually 0xbf000000),
it is considered to be in the kernel's address space, not the
process', so it doesn't show up in /proc/<pid>/pagemap.
When the vector page is detected, using the name "[vectors]",
remove it from the map.
Change-Id: I5f0758bbe5d2b927056fa9fee684fea63dd0fa4b
|
|
pm_process_usage can return an error, and leave procs[i]->usage
unitialized. Detect the error case and print a warning. Also
make the initialization of procs[i]->usage to 0 unconditional,
so nothing will be printed in the final procrank stats when an
error is detected.
Change-Id: I03f90ae6a5ebb201b6e9e43593cec225e1a8ded0
|
|
|
|
Bug: 5010576
Change-Id: I5a7df94d99dcb6f1de98bd9b8f659fd51a491c82
|
|
|
|
Change-Id: Ie6b25393bfd61be54dc6b58b81bffbc2d26b40f5
|
|
Change-Id: Ica273bf302bbc5025e48aa62eff5bf64fbdcac77
|
|
Change-Id: Ia4d178b535e1c154cf2b8e826ea1fba342b05658
|
|
Fix definition of uuid struct.
Properly set the block_group number in each backup copy of the superblock.
Adjust the info.total_inodes field after rounding up the inodes per block group.
Add the option -t to initialize all the inode tables. If also specified with
the -s option, use the CHUNK_TYPE_FILL sparse records to initialize them.
Change-Id: Idd8bcee1b9bde3e82ad8da89ef974fbc12d7a6c6
|
|
Add debugging test code to specify where to bail partway through
Add a script to drive the test code for automated testing. This
found 4 bugs!
Change-Id: I14dc8b1e2c9d2d2f332346958d488feaf768d396
|
|
Add check for filesystem that needs the journal run
Fix check for a cleanly unmounted filesystem.
Fix computation of new inode number
Fix check in two places for high bit being set on inode num
Only run sanity check pass if filesystem fixup state is unset
(i.e. not partway through the conversion)
Change-Id: Ib5521e7f3c04d32c02d9890644a99378e3a3659e
|
|
The filesystems made by make_ext4fs for the Gingerbread and initial Honeycomb
releases did not round up the number of inodes/block_group to use all the
space available. This worked, but newer kernels have additional code that
refuses to write to these filesystems. There is a kernel patch available
to fix the problem in the kernel, but this tool was also written to fixup
the filesystems to be compliant with the new kernels.
Change-Id: I317b02ee2132ce6a0daeb4d1b40ad5ef57a0fb32
|
|
Based on code review comments when the code was moved from
crespo specific code to ext4 generic code.
Change-Id: Ia09261563863f6db1d5eadb5a3677b0b38382b93
|
|
Prime also uses this, so it's no longer crespo specfic.
Change-Id: Idd35ee85599717a06902f696a73e5a4bd6522ac6
|
|
Change-Id: Ic2b2fe51f052e8b2a2d04f058354490e9cea07ae
|
|
|
|
Kernel version 2.6.37 and later refuse to write to filesystems
for which inodes per group is not divisible by inodes per block.
Round the inodes per group up.
Change-Id: I9eb4fb5dd616a8abfd2f1fdc4b7d05538a2fc8fe
|
|
Add a commandline argument to provide bpp.
test-fb-simple <bpp> <w> <h> <buffer#>
Valid values for bpp are 16 and 32.
Change-Id: I60485700c379746948abda6f03d617d9f605fb7b
Signed-off-by: Dima Zavin <dima@android.com>
|
|
When computing the size of a directory that is being created,
make sure to include the size of the null entry at the end of
the list of entries in that directory.
Change-Id: I7deb05a170911cd0cc8517600c9747f64915606f
|
|
killling recovery
* commit '2ae7663e1e064000356ee9e49ebd08bdddc545f9':
Implement poor man's exception handling to prevent errors from killling recovery
|
|
make_ext4fs was originally designed to be a standalone program, and the
error() macro calls exit() after printing a diagnostic. However, when
it was turned into a library that recovery links against, calling exit()
kills the entire recovery process, which the kernel starts back up,
which dies, and the cycle continues.
This changes make_ext4fs to use setjmp() early, and changes error to
use longjmp to return to that point. When longjmp is called, the
error handler just returns an error. If it was invoked as a standalone
program, this causes the program to exit with an error. If it was
invoked as a library function, the error return is gracefully handled
by recovery.
Change-Id: Iad50f46cac0ac399b5cdfd083d5712f88a10e48e
|
|
|
|
* commit '8c00dae657ce7941121ec218d07adeab9ee3d8a1':
Make a failure of the BLKDISCARD ioctl non-fatal.
|
|
gigabytes.
* commit 'e2e962fdf6fa3d29986b92438e2f9c2b8e9d2284':
Fix to properly wipe filesystems larger than 4 gigabytes.
|
|
* commit '427c3a2177eb4d34c88406bb86107548737168cc':
Make a failure of the BLKDISCARD ioctl non-fatal.
|
|
* commit '3e620597671c7efa678c1cd804202cff513bc751':
Fix to properly wipe filesystems larger than 4 gigabytes.
|
|
Failure to erase the partition before making a new filesystem
should not be fatal, especially since this is called from recovery,
and if make_ext4fs() dies, recovery dies.
Change-Id: I1b50b8c486d403a5d9cbd5e9de73cf752d8ef8f7
|
|
Pass the wipe size to the wipe function in a 64 bit int.
Change-Id: Ia6a0c70c8ddacaae35d17732d61b126219e5c158
|
|
Change-Id: Iafd936bcb30e3fcc18c522a701d567dd744f2fa3
Signed-off-by: Iliyan Malchev <malchev@google.com>
|
|
Pass the wipe size to the wipe function in a 64 bit int.
Change-Id: Iac95693efa2d9cadbe77aa83b5ff01d7fc8d8b70
|
|
The X86 target couldn't compile some tests that use assembly, but it can
compile other things that now depend on some projects in the test subdir
(namely libtestUtil for OpenGL tests)
Also fatblock was including <sys/endian.h> directly instead of
<endian.h> which broke some stuff on X86
Change-Id: Ib71a6e548189ca2eacfd0caa7dbd17d200efd0ea
|
|
|
|
|
|
|
|
|