Unfortunately there are no unit test cases; at the end of the day, this is often a part of tuning the platform and seeing that it is behaving appropriately in use.
That said, I would really like to have some CTS tests for this. These are tricky to do since the guarantees are so loose, but I think some basic stuff can be done. For example a test could: 1. Start an activity in process A. 2. Start a service in process B. 3. Start another activity in process C. Now process A is a background process, B is a service, and C is foreground. If C starts using memory, A should be killed before B which is killed before C. It is tricky however to put this together so it works as a unit test with dependable behavior. On Wed, Nov 24, 2010 at 10:37 PM, Raja Pavan <rajain...@gmail.com> wrote: > Is there any test methodology available from android, for testing LMK > driver? > > On Nov 25, 11:17 am, Dianne Hackborn <hack...@android.com> wrote: > > There are no guarantees about things being killed at an exact level. In > > fact what the current "memory level" is, is quite fuzzy if not just > outright > > not knowable. > > > > Is there a problem you are trying to address? > > > > > > > > > > > > On Wed, Nov 24, 2010 at 9:49 PM, Raja Pavan <rajain...@gmail.com> wrote: > > > Hi Dianne > > > > > I did run the test application (shared with you) on emulator. > > > > > It was killed. > > > > > It has been observed that the adb shell is also killed (as it is not > > > available anymore) > > > > > But How do I know whether the process is crashing at the watermark > > > level? > > > > > Thanks > > > Raja > > > > > On Nov 23, 10:19 pm, Dianne Hackborn <hack...@android.com> wrote: > > > > Have you run your tests on the stock platform (even just the > emulator) to > > > > see if you are getting the behavior you expect there? There are a > *lot* > > > of > > > > potential gotchas if trying to test this behavior: running at the > right > > > > oom_adj, fuzziness in the definition of "free memory", etc. I would > > > suggest > > > > validating your test first on a known working build before using it > to > > > draw > > > > conclusions about the behavior of your own build. > > > > > > On Tue, Nov 23, 2010 at 4:45 AM, Raja Pavan <rajain...@gmail.com> > wrote: > > > > > We are trying to test Android LowMemoryKiller Driver using a sample > > > > > test application. > > > > > > > /sys/module/lowmemorykiller/parameters/adj 0,1,2,7,14,15 > > > > > > > MINFREE module parameter is set from the application. > > > > > > > echo 1536,2048,4096,5120,5632,6144 >/sys/module/lowmemorykiller/ > > > > > parameters/minfree > > > > > > > The test application is pasted at the end of this mail. > (lowmemTest.c) > > > > > > > Expected Behavior: We have set the oom_adj of the lowmemTest.c to > 13. > > > > > It is expected to get killed when the free memory in the system (as > > > > > seen from the putput of free() command is 5120*4*1024 = 20MB . > > > > > > > We are printing the free memoy of the system by invoking free() > inside > > > > > the lowmemoryTest.c > > > > > > > Observed Behavior: > > > > > (1) But the process is not being killed at the 20MB free memory. > > > > > (2) This process is getting killed at the free memory close to the > > > > > min_free_kbytes (/proc/sys/vm/min_free_kbytes) > > > > > > > Need Support on these Isssues (Any Patch to kernel) > > > > > > > (1) Is there any patch to the linux 2.6.30 kernel (MM or some other > > > > > subsystem) for getting the desired behavior from lowmemorykiller > > > > > driver. > > > > > (2) We have taken the lowmemory killer for this kernel from android > > > > > git, will this work? > > > > > (3) Is there any test application available to test the > > > > > lowmemorykiller, provided by Google? > > > > > (4) Can some provide us clue about improved debugging of this > driver? > > > > > > > Please extend help ASAP. > > > > > > > Thanks > > > > > Raja > > > > > > > -- > > > > > unsubscribe: > > > > > android-kernel+unsubscr...@googlegroups.com<android-kernel%2bunsubscr...@googlegroups.com> > <android-kernel%2bunsubscr...@googlegroups.com> > > > <android-kernel%2bunsubscr...@googlegroups.com> > > > > > website:http://groups.google.com/group/android-kernel > > > > > > -- > > > > Dianne Hackborn > > > > Android framework engineer > > > > hack...@android.com > > > > > > Note: please don't send private questions to me, as I don't have time > to > > > > provide private support, and so won't reply to such e-mails. All > such > > > > questions should be posted on public forums, where I and others can > see > > > and > > > > answer them.- Hide quoted text - > > > > > > - Show quoted text - > > > > > -- > > > unsubscribe: > > > android-kernel+unsubscr...@googlegroups.com<android-kernel%2bunsubscr...@googlegroups.com> > <android-kernel%2bunsubscr...@googlegroups.com> > > > website:http://groups.google.com/group/android-kernel > > > > -- > > Dianne Hackborn > > Android framework engineer > > hack...@android.com > > > > Note: please don't send private questions to me, as I don't have time to > > provide private support, and so won't reply to such e-mails. All such > > questions should be posted on public forums, where I and others can see > and > > answer them.- Hide quoted text - > > > > - Show quoted text - > > -- > unsubscribe: > android-kernel+unsubscr...@googlegroups.com<android-kernel%2bunsubscr...@googlegroups.com> > website: http://groups.google.com/group/android-kernel > -- Dianne Hackborn Android framework engineer hack...@android.com Note: please don't send private questions to me, as I don't have time to provide private support, and so won't reply to such e-mails. All such questions should be posted on public forums, where I and others can see and answer them. -- unsubscribe: android-kernel+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-kernel
