[PATCH v5 8/8] add a new interface to show the memory usage of 1st kernel

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello Baoquan,

>On 08/25/14 at 04:04pm, Vivek Goyal wrote:
>> On Mon, Aug 25, 2014 at 02:36:22PM +0800, Baoquan He wrote:
>
>> > The print is like below:
>> >
>> > ~$ ./makedumpfile --mem-usage /proc/kcore
>> > The kernel version is not supported.
>> > The created dumpfile may be incomplete.
>>
>> I still think that above messages should go. It does not make
>> any sense with --mem-usage. No dumpfile is being created here.
>
>This is printed by function get_kernel_version(), it will be used by
>other makedumpfile main code flows too. This message is used to warn
>users that users' using kernel may not be tested suffciently. I think
>those tests are mainly taken by Atsushi. When he finished sufficient
>tests on a new version of kernel, the LATEST_VERSION will be changed to
>be a larger value. I am fine with this message, since it won't occur on
>our distribution, anyway kernel for distribution are all tested. And if
>LATEST_VERSION is older than our distribution kernel, maintainer may be
>pushed to take tests and update this value. So don't worry about this
>message.
>
>But I would like to change the message like below to clean up
>misunderstanding if Atsushi doesn't object, this mostly occur when
>people working on latest upstream kernel.
>-------------------------------------------------
>The kernel version is not supported.
>The makedumpfile operation may not be successful.
>--------------------------------------------------

Good change, let's do it. --mem-usage doesn't create a dumpfile
but the behavior still depend on the kernel version, this message
is meaningful. Also, as Vivek said, the latter line can be
changed based on the operating mode, but I don't think it's
necessary.

>>
>> > Excluding unnecessary pages        : [100.0 %] |
>>
>>
>> >
>> > Page number of memory in different use
>> > --------------------------------------------------
>>
>> I think above header can completely go away.  It kind of looks odd.
>
>OK, will remove it.
>
>>
>> > TYPE		PAGES			EXCLUDABLE	DESCRIPTION
>> > ZERO		29149           	yes		Pages filled with zero
>> > CACHE		171288          	yes		Cache pages
>> > CACHE_PRIVATE	12051           	yes		Cache pages + private
>> > USER		31816           	yes		User process pages
>> > FREE		3700059         	yes		Free pages
>> > KERN_DATA	105305          	no		Dumpable kernel data
>> >
>> > Total pages on system:	4049668
>> >
>> > Showing page number of memory in different use successfully.
>>
>> I think we don't need above line. I am not even sure what does it mean.
>
>Will remove it.
>
>> >
>> > makedumpfile Completed.
>>
>> We don't need above line either.
>
>This is also a public message printing, means if makedumpfile operation
>is successful or not. Maybe I can add a check like below:
>
>@@ -9546,7 +9553,7 @@ main(int argc, char *argv[])
>        retcd = COMPLETED;
> out:
>        MSG("\n");
>-       if (retcd == COMPLETED)
>+       if ((retcd == COMPLETED) && (!info->flag_mem_usage))
>                MSG("makedumpfile Completed.\n");
>        else
>                MSG("makedumpfile Failed.\n");

This code always show "makedumpfile Failed" for --mem-usage :-)

Anyway,

>Hi Atsushi,
>
>How do you think about this? And for the excluding progress indication
>too, add a check that if it's mem-usage handling, not printing.

I don't think we need to change the both messages, they indicate
the actual progress and result even in the mem-usage mode.


Thanks
Atsushi Kumagai

>Thanks
>Baoquan
>
>>
>> In output we should just show the actual table. If there is an error,
>> we should output error and exit (no table).
>
>>
>> Thanks
>> Vivek
>>
>> _______________________________________________
>> kexec mailing list
>> kexec at lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/kexec



[Index of Archives]     [LM Sensors]     [Linux Sound]     [ALSA Users]     [ALSA Devel]     [Linux Audio Users]     [Linux Media]     [Kernel]     [Gimp]     [Yosemite News]     [Linux Media]

  Powered by Linux