esp32/mpi: Added alternate workaround for MPI data corruption issue

- Use DPORT_WRITE_REG (volatile writes) wrappers to write to the
  MPI peripheral
- Updated the previous workaround added for the same issue as it
  was failing in some long runs and with `COMPILER_OPTIMIZATION_PERF`
  enabled.
- The test performance numbers had to be updated due to the
  performance penalty introduced by this fix.

Closes https://github.com/espressif/esp-idf/issues/10403

components/idf_test/include/esp32/idf_performance_target.h

@@ -22,11 +22,11 @@
 #define IDF_PERFORMANCE_MAX_TIME_SHA512_32KB                                    4500
 
 #define IDF_PERFORMANCE_MAX_RSA_2048KEY_PUBLIC_OP                               19000
-#define IDF_PERFORMANCE_MAX_RSA_2048KEY_PRIVATE_OP                              420000
+#define IDF_PERFORMANCE_MAX_RSA_2048KEY_PRIVATE_OP                              450000
 #define IDF_PERFORMANCE_MAX_RSA_3072KEY_PUBLIC_OP                               33000
 #define IDF_PERFORMANCE_MAX_RSA_3072KEY_PRIVATE_OP                              950000
 #define IDF_PERFORMANCE_MAX_RSA_4096KEY_PUBLIC_OP                               90000
-#define IDF_PERFORMANCE_MAX_RSA_4096KEY_PRIVATE_OP                              1700000
+#define IDF_PERFORMANCE_MAX_RSA_4096KEY_PRIVATE_OP                              1900000
 
 #define IDF_PERFORMANCE_MAX_SPI_PER_TRANS_POLLING                               15
 #define IDF_PERFORMANCE_MAX_SPI_PER_TRANS_POLLING_NO_DMA                        15

components/mbedtls/port/esp32/bignum.c

@@ -69,22 +69,22 @@ void esp_mpi_interrupt_clear( void )
 */
 
 /* Please see detailed note inside the function body below.
- * Relevant: https://github.com/espressif/esp-idf/issues/8710 and IDF-6029
+ * Relevant: IDF-6029
+             https://github.com/espressif/esp-idf/issues/8710
+             https://github.com/espressif/esp-idf/issues/10403
  */
-static inline void __attribute__((optimize("-fno-tree-loop-distribute-patterns")))
-            mpi_to_mem_block(uint32_t mem_base, const mbedtls_mpi *mpi, size_t hw_words)
+static inline void mpi_to_mem_block(uint32_t mem_base, const mbedtls_mpi *mpi, size_t hw_words)
 {
-    uint32_t *pbase = (uint32_t *)mem_base;
     uint32_t copy_words = MIN(hw_words, mpi->MBEDTLS_PRIVATE(n));
 
     /* Copy MPI data to memory block registers */
     for (uint32_t i = 0; i < copy_words; i++) {
-        pbase[i] = mpi->MBEDTLS_PRIVATE(p[i]);
+        DPORT_REG_WRITE(mem_base + i * 4, mpi->MBEDTLS_PRIVATE(p[i]));
     }
 
     /* Zero any remaining memory block data */
     for (uint32_t i = copy_words; i < hw_words; i++) {
-        pbase[i] = 0;
+        DPORT_REG_WRITE(mem_base + i * 4, 0);
     }
 
 #if _INTERNAL_DEBUG_PURPOSE
@@ -94,14 +94,9 @@ static inline void __attribute__((optimize("-fno-tree-loop-distribute-patterns")
      * specific write (store) operation to the MPI peripheral block was getting lost erroneously.
      * Following data re-verify loop could catch it during runtime.
      *
-     * As a workaround, we are disabling loop distribute patterns for this function and hence
-     * compiler does not enforce usage of `memset` (or `memcpy`) calls for this routine. It
-     * appears that `-ftree-loop-distribute-patterns` was enabled with O2/Os starting from
-     * GCC-10.x. It is quite possible that there is some issue with DPORT write with sequence of
-     * store instructions as generated by `memset` call, but for now this should serve as good
-     * interim workaround without any impact on the performance.
+     * As a workaround, we are using DPORT_WRITE_REG (volatile writes) wrappers to write to
+     * the MPI peripheral.
      *
-     * Please see IDF-6029 for more details.
      */
 
     //for (uint32_t i = copy_words; i < hw_words; i++) { assert(pbase[i] == 0); }