yamt commented on PR #13222:
URL: https://github.com/apache/nuttx/pull/13222#issuecomment-2317752875

   
   > > * [Increase the chance for _assert to work early in the boot 
#13233](https://github.com/apache/nuttx/pull/13233)
   > 
   > Although this change makes sense, I didn't find any issues related direct 
to it.
   
   it prevents DIAGASSERT on CONFIG_MM_REGIONS from showing the panic messages.
   
   > > * for some reason, the last page is corrupted.
   > > 
   > > the following is with CONFIG_MM_FILL_ALLOCATIONS=y. thus i expect it's 
all 0xcc. but as you can see there are some 32byte patterns. (32byte because of 
EVENODD mode i guess) with NORMAL mode, the symptom disappears. but i guess 
NORMAL mode doen't meet the general coherency expectations for nuttx smp. do 
you have any ideas?
   > 
   > I have no idea. After increasing the `CONFIG_MM_REGIONS`, it was able to 
boot, but `free` command fails:
   > 
   > ```
   > [CPU0] _assert: Assertion failed info.uordblks + info.fordblks == 
info.arena: at file: mm_heap/mm_mallinfo.c:157
   > ```
   > 
   > In fact, if we disable `CONFIG_DEBUG_ASSERTIONS`, `free` returns:
   > 
   > ```
   > nsh> free
   >                  total       used       free    maxused    maxfree  nused  
nfree
   > esp32-imem:      98300       8628      89672       9024      89672      5  
    1
   >       Umem:    4402076    4263516    4398464       4008    4194288     27  
    5
   > ```
   > 
   > It isn't evaluating the used/free size correctly.
   
   it's because the last page corruption.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to