[RTEMS Project] #3007: ARM caching issues

[RTEMS trac]

#3007: ARM caching issues
-------------------------------+------------------------------
 Reporter:  munster            |       Owner:  joel.sherrill@…
     Type:  defect             |      Status:  new
 Priority:  normal             |   Milestone:  4.12
Component:  bsps               |     Version:  4.12
 Severity:  normal             |  Resolution:
 Keywords:  ARM,cache,L2C 310  |
-------------------------------+------------------------------

Comment (by Sebastian Huber):

 Replying to [comment:5 munster]:
 > Replying to [comment:4 Sebastian Huber]:
 > > Please don't change the function signatures. Please generate one patch
 for each issue.
 > I have attached separate patches for these two issues.

 Could you please format the patches via "git format-patch":

 https://devel.rtems.org/wiki/Developer/Contributing

 Please help to improve the wiki and website if something is unclear or
 hard to find.

 >
 > > Its up to the user to ensure that a cache line invalidation is safe.
 The cache manager should not take care of this.
 > Generally you are right. However, the user may not be aware of this
 particular problem.

 The we should fix this issue with better documentation. In

 https://git.rtems.org/rtems/tree/cpukit/rtems/include/rtems/rtems/cache.h#n111

 we have
 {{{
 /**
  * @brief Invalidates multiple data cache lines.
  *
  * The cache lines covering the area are marked as invalid.  A later read
  * access in the area will load the data from memory.
  *
  * In case the area is not aligned on cache line boundaries, then this
  * operation may destroy unrelated data.
  *
  * @param[in] addr The start address of the area to invalidate.
  * @param[in] size The size in bytes of the area to invalidate.
  */
 void rtems_cache_invalidate_multiple_data_lines(
   const void *addr,
   size_t size
 );
 }}}

 The cache manager documentation probably needs to move into the user
 manual. Maybe some examples help to show what can go wrong.

--
Ticket URL: <http://devel.rtems.org/ticket/3007#comment:6>
RTEMS Project <http://www.rtems.org/>
RTEMS Project