[VirtLayer] Midterm patches and blog update

Gedare Bloom gedare at rtems.org
Mon Aug 19 18:48:10 UTC 2013

I think you can use the rtems-testing.git module with the pc386 script
to run the tmtests in non-interactive mode. Then you just need to run
all of them (easy to script), and it should dump the output to the
"log/" subdirectory. I think there are instructions on our wiki page
about Qemu... ask if you don't figure it out easily. :)


On Mon, Aug 19, 2013 at 12:34 PM, Philipp Eppelt
<philipp.eppelt at mailbox.tu-dresden.de> wrote:
> Is there an automatic way to do all tmtests on qemu?
> The score-libcpu_native patch is ready. I just need to do the tests.
> On 08/15/2013 06:25 PM, Joel Sherrill wrote:
>> My experience in the past is that it is accurate enough to see changes. It is more about the relative change that precise numbers.
>> Gedare Bloom <gedare at rtems.org> wrote:
>> You ought to test the patch that moves code from score to libcpu with
>> the tmtests to ensure that none of the timing changes. (Hm, does
>> pc386/qemu offer accurate tmtest timing results?)
>> -Gedare
>> On Wed, Aug 14, 2013 at 1:36 PM, Gedare Bloom <gedare at rtems.org> wrote:
>>> Please merge the part of the libcpu_i386_midterm.patch that contains
>>> the relocated code into the score_i386_midterm.patch since the score
>>> patch will not work right without reimplementing the libcpu part. Make
>>> the addition of the virtual subdirectory of the libcpu a separate
>>> patch that includes everything needed for it to work, probably by
>>> merging it with the virtPok BSP.
>>> Please try to submit the patches directly to the mailing list with
>>> git-send-email, it is easier to review patches that are emailed
>>> directly instead of attached. Instructions can be found on the RTEMS
>>> wiki Git_Users page.
>>> Other notes:
>>> * avoid two or more blank lines in a row
>>> * break lines longer than 80 characters
>>> * don't use "Intel" to specify the i386 type in comments (other x86
>>> brands should work..)
>>> * put the extern "C" after include files. "C" files should not need
>>> it, only header files.
>>> * Delete "Formerly contained in and extracted from libcpu/i386/cpu.h"
>>> from interrupts.h header comment. Possibly add comment "Relocated from
>>> score/cpu/i386"
>>> * rename files to avoid humpBackNotation. Files in RTEMS are usually
>>> alllowercase so let us follow the usual way.
>>> * Remove "Date: XX/YY/ZZZZ" from file comments
>>> * Replace "Author: Philipp Eppelt" with a copyright statement e.g.
>>> "Copyright (c) 2013 Philipp Eppelt."
>>> * "virt_disableInterrupts" and "virt_enableInterrupts" are undefined.
>>> Also they should be given a proper RTEMS name, e.g.
>>> _CPU_Virtual_Disable_interrupts() or something.
>>> * Add your copyright to the virtual/include/rtems/score/interrupts.h.
>>> You substantially transformed this code and should lay claim to your
>>> copyright.
>>> -Gedare
>>> On Wed, Jul 31, 2013 at 4:17 PM, Philipp Eppelt
>>> <philipp.eppelt at mailbox.tu-dresden.de> wrote:
>>>> Patch for libcpu, introducing CPU models 'native' and 'virtual' and
>>>> implementing functions removed from score.
>>>> _______________________________________________
>>>> rtems-devel mailing list
>>>> rtems-devel at rtems.org
>>>> http://www.rtems.org/mailman/listinfo/rtems-devel
>> _______________________________________________
>> rtems-devel mailing list
>> rtems-devel at rtems.org
>> http://www.rtems.org/mailman/listinfo/rtems-devel

More information about the devel mailing list