This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: SystemTap for Android - patchset


On 07.07.2016 20:47, David Smith wrote:
> On 07/06/2016 07:29 AM, Alexander Lochmann wrote:
>> 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.
> 
> The problem you are trying to fix in your patch "[PATCH 1/3] Definition
> of cputime_to_usecs in Linux kernel 3.0 is broken" is interesting. I
> don't think you've got quite the right solution. Testing for a kernel
> version here works for you, but really isn't a general solution -
> depending on arch there could be a kernel in that range with a
> reasonable cputime_to_usecs().
I see. :-/ Lesson learned.
> 
> ====
> diff --git a/tapset/linux/task_time.stp b/tapset/linux/task_time.stp
> index f86f984..f3c276c 100644
> --- a/tapset/linux/task_time.stp
> +++ b/tapset/linux/task_time.stp
> @@ -27,8 +27,12 @@
>   * Yet note some kernels (RHEL6) may already have both...  */
>  #if defined(cputime_to_usecs)
>  #if !defined(cputime_to_msecs)
> +#if LINUX_VERSION_CODE <= KERNEL_VERSION(3,0,200) && LINUX_VERSION_CODE
>> = KERNEL_VERSION(3,0,0)
> +#define cputime_to_msecs(__ct)		_stp_div64(NULL,
> ({cputime_to_usecs(__ct)}), 1000ULL)
> +#else
>  #define cputime_to_msecs(__ct)		_stp_div64(NULL,
> cputime_to_usecs(__ct), 1000ULL)
>  #endif
> +#endif
> 
>  /* Kernels before 2.6.37 have cputime_to_msecs, but not usecs. */
>  #elif defined(cputime_to_msecs)
> ====
> 
> I've got a couple of thoughts here:
> 
> - I wonder if we just couldn't always use your workaround (along with a
> comment about why we're doing that).
Why not? :) Since the kernel and every module of course is compiled with
-O3 (or -O2?),  it shouldn't make any difference in the resulting
assembler code.
> 
> - Here's an untested workaround idea that might work for all kernels:
> 
> #define cputime_to_msecs(__ct)	 		 		\
> 	({							\
> 		unsigned long __usecs = cputime_to_usecs(__ct);	\
> 		_stp_div64(NULL, __usecs, 1000ULL);		\
> 	})
> 
And so does this version. The compiler should be smart enough to
recognize __usecs is a local variable, and thus could be stored in a
register.

- 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

Attachment: signature.asc
Description: OpenPGP digital signature


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]