RE: Some Problem come out during runing systemtap module on Android phone.

[ììíB <jaehyek dot choi at lge dot com>]

Hi, Frederic.

Thanks for your answer.

After your mail, I have tried to find why systemtap on gingerbread phone is not working.
After all, I got a different point between the several kernel versions which were supplied  by qualcomm,
and I verified my systemtap works fine.

Below is  before, after and original  from the file name "arch/arm/kernel/vmlinux.lds.s, around 99 line" .
>> Before. ( got the problem .)

			*(.gnu.warning)
			*(.rodata)
			*(.rodata.*)
			*(.glue_7)
			*(.glue_7t)
		*(.got)			/* Global offset table		*/
	}
#ifdef CONFIG_STRICT_MEMORY_RWX
	. = ALIGN(1<<SECTION_SHIFT);
#endif

	RO_DATA(PAGE_SIZE)

>> after ( fixed the problem )

			*(.gnu.warning)
			*(.rodata)
			*(.rodata.*)
#ifdef CONFIG_STRICT_MEMORY_RWX
            . = ALIGN(1<<SECTION_SHIFT);
#endif
			*(.glue_7)
			*(.glue_7t)
		*(.got)			/* Global offset table		*/
	}

	RO_DATA(PAGE_SIZE)

>> original kernel version 

			*(.gnu.warning)
			*(.rodata)
			*(.rodata.*)
			*(.glue_7)
			*(.glue_7t)
		*(.got)			/* Global offset table		*/
	}

	RO_DATA(PAGE_SIZE)


Accordinng the above ,   what caused the problem is  the position of ". = ALIGN(1<<SECTION_SHIFT);", 
and the original version  doesnât have the line.

On your guesswork,  why do you think the qualcomm  inserts the line?


.Other questions.

I want to monitor the CPU time occupation of some process, 
 and also to monitor the reduction of kernel memory which is caused from the abnormal memory usage of some process .

could you share some systemtap codes to me about the above two concern ?
or recommend some samples  among the distribution of systemtap ? 



Best Regars
Jaehyek Choi


-----Original Message-----
From: systemtap-owner@sourceware.org [mailto:systemtap-owner@sourceware.org] On Behalf Of Turgis, Frederic
Sent: Thursday, September 01, 2011 5:38 PM
To: ììíB
Cc: 'Josh Stone'; systemtap@sourceware.org
Subject: RE: Some Problem come out during runing systemtap module on Android phone.

> But the target âandroid gingerbread phoneâ is not working.
> when I run the âhelloworld.stpâ,  an abnormal message comes
> out like below.
>
> /data/staphome/bin# ./staprun helloworld.ko helloworld
> /data/staphome/bin/staprun: No such file or directory (  <-
> abnormal message )
Try first ./staprun to see if "help" is coming out. "printf" shall work ! Try to chmod staprun and stapio, this does not look related to systemtap
I made systemtap work on Froyo and Gingerbread, well this is essentially a kernel related tool on ARM (no time to investigate/code utrace/uprobes on our side :-( ) so it shall work on any pastry, as google does not change that much the kernel. But userspace is killing me as a developer.


>
>
> Also when I run the â ./staprun iotop.koâ on target phone,
> the phone is gone to reset .
> Could you help me to resolve this problem ?
Mmmmh, we have hit issues when running several scripts in parallel (on Android, works fine on Ubuntu). Works first time, not 2nd time. Or issues when probing all system calls. But iotop.stp is working fine on my Gingerbread (K2.6.35)

>
> As your reference,
> . the Froyo phone is including the kernel version 2.6.32
> which was built in ubuntu 32bit, . the Gingerbread phone is
> including the kernel version 2.6.35.11 which was built in
> ubuntu 64bit.
We are at kernel 2.6.35. I don't really track the potential incompabilities between kernels and systemtap. We had to modify ptrace.h in the past, I don't remember when we started no longer needing it

> . And both phone is not OMAP-based . ( Should I use the OMAP
> platform ? )
You don't need an OMAP platform. We don't make any specific change for OMAP. Using /data/staphome instead of /home/xxx is totally OK as far as you respect paths on desktop and target, which is the case for you.

Texas Instruments France SA, 821 Avenue Jack Kilby, 06270 Villeneuve Loubet. 036 420 040 R.C.S Antibes. Capital de EUR 753.920