SystemTap for Android - patchset

[Alexander Lochmann]

So. Let me start.
First of all, I extracted the patches properly, and attached the files.
(Btw, I found a third bug. :) )
FYI, I just fixed the bugs for the kernel versions I'm dealing with,
because I don't know which other versions are affected as well.

On 01.07.2016 19:46, Josh Stone wrote:
> On 07/01/2016 09:15 AM, Alexander Lochmann wrote:
>>
>> Hi folks!
>>
>> Finally, I decided to submit my patch, which makes SystemTap work for
>> Android. Moreover, it adds two new features:
>> - Support for ignoring all available tapset directories, except the one
>> that is provided by -K
> 
> Can you explain why you need this?
> 
> One can already override the primary tapset directory with the
> environment SYSTEMTAP_TAPSET.  Ignoring -I options also seems
> questionable, especially since this depends on the order -- it looks
> like -I specified after -K will still be used, while those before will
> be ignored.
> 
Honestly? I don't know. :D I had some trouble with SystemTap looking for
tapsets systemwide, which made me write that patch/fix. It's been a long
time since I wrote it. Anyway, I tried your proposed solution, and it
works. I'll adapt my build script to use the enviroment variable.
You can skip that part of the patch if you want.
>> - Support for a pid file in staprun, parameter is -U
> 
> And you added -M on stap itself.  I don't understand either of those
> letter choices.
Did I? It might be possible that I want be able to pass the argument to
staprun.

I chose the letters randomly. For me, it doesn't matter. I just need an
argument to tell staprun to create a pid file. :)
> 
> The functionality itself is not too controversial, but I'd still like to
> see your justification in the commit log, and some examples how this is
> used in Android.
Yeah, sure. Since I run stap on Android, I have a background service,
which periodically checks if every single stap instance is still
running. Yes, it might be possible having more than one stap script
running. :)
Moreover, if I start it more than once, I want to be able to kill
specific instance. For that I need its process id.

- Alex
> 
>> I had to modify several source files of staprun. Those changes are
>> mostly copied from the corresponding files contained in commit
>> 2c10863bfe41b51272eff714a837f4977bdc257a. For some reasons, those ifdef
>> parts have been removed. I readded them, and changed the macro, which
>> activates them.
> 
> That commit is "man stap: fixed typos".  Did you mean something else?
> 
>> The patch contains two bugfixes for the SystemTap as well.
>> Unfortunately, I failed to extract those fixes properly. :(
> 
> As David said, it would help us a lot if you separated each of these
> changes into distinct commits.  If you have the changes in a working git
> directory, "git add -p" can help you mark specific changes to be
> committed.  "git-gui" is also pretty good for selecting hunks.  Feel
> free to ask for help on #systemtap if you still have trouble.
> 
>> The first fix starts at line 510, and goes until line 555.
>> Since an older kernel like 3.0 does not support uprobes, systemtap
>> includes 'runtime/linux/task_finder_stubs.c'. That file itself does
>> *not* include 'syscall.h', which declares several syscall-related functions.
>> The second fix starts at line 1106. For some reasons in the Linux kernel
>> 3.0 the macro cputime_to_usecs() has a semicolon at the end of its
>> definition. Therefore, the defition of cputime_to_msecs() in '
>> tapset/linux/task_time.stp' must be modified to deal with that fact.
>>
>> Cheers,
>> Alex
>>
>> ---
>> Technische Universität Dortmund
>> Alexander Lochmann                PGP key: 0xBC3EF6FD
>> Otto-Hahn-Str. 16                 phone:  +49.231.7556141
>> D-44227 Dortmund                  fax:    +49.231.7556116
>> http://ess.cs.tu-dortmund.de/Staff/al
>>
> 


-- 
Technische Universität Dortmund
Alexander Lochmann                PGP key: 0xBC3EF6FD
Otto-Hahn-Str. 16                 phone:  +49.231.7556141
D-44227 Dortmund                  fax:    +49.231.7556116
http://ess.cs.tu-dortmund.de/Staff/al
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Definition-of-cputime_to_usecs-in-Linux-kernel-3.0-i.patch
Type: text/x-patch
Size: 1204 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/systemtap/attachments/20160706/7f66e051/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-The-syscall-defines-were-not-compatible-with-older-k.patch
Type: text/x-patch
Size: 2527 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/systemtap/attachments/20160706/7f66e051/attachment-0001.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0003-Linux-kernel-3.4-does-not-define-PR_SET_MM_ARG_START.patch
Type: text/x-patch
Size: 1035 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/systemtap/attachments/20160706/7f66e051/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://sourceware.org/pipermail/systemtap/attachments/20160706/7f66e051/attachment.sig>