[GitHub] [incubator-nuttx] pussuw commented on pull request #6134: RISC-V: Fix system crash when FPU is in use

[GitBox <gi...@apache.org>]

pussuw commented on PR #6134:
URL: https://github.com/apache/incubator-nuttx/pull/6134#issuecomment-1301895125

   @masayuki2009 It has been a while but I was revisiting the context saving code and remembered you had an issue with FPU on QEMU. I found at least one issue, that might be the cause for my crashes:
   
   The int/fpu register restoration ordering in exception_common:
   ```
     load_ctx   sp
     riscv_loadfpu sp
   ```
   is bad, because riscv_loadfpu uses registers t0,t1,t2
   ```
   .macro riscv_loadfpu out
   
   #ifdef CONFIG_ARCH_FPU
     REGLOAD      t0, REG_INT_CTX(\out)
     li           t1, MSTATUS_FS
     and          t2, t0, t1
     li           t1, MSTATUS_FS_INIT
     ble          t2, t1, skip_load_fpu
   ```
   
   So the bad ordering restores the original integer registers including t0,t1,t2, and then these are immediately destroyed by riscv_loadfpu.
   
   One potential fix would be use the "bad" ordering above, but to restore t0,t1,t2 from the context after the FPU context is restored:
   
   ```
   skip_load_fpu:
   ... restore t0,t1,t2
   ```
   
   I will try this again when I have time, but if you ever get the urge to get FPU working on QEMU this might be a potential fix.


-- 
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: commits-unsubscribe@nuttx.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org