Some assembly required... seeing trap when using ee.vld.128.ip instruction
Posted: Mon Oct 17, 2022 11:09 am
I am seeing a trap when I use the ee.vld.128.ip instruction on an ESP32 S3. As far as I can tell, the way I'm using it should be fine. The value of a12 before this instruction points to a valid memory location: 0x40378d40. The trap also says that the address being accessed is invalid, though the address can be accessed just fine in the preceding l32i instruction which I put in to verify I didn't screw anything up. While I have extensive assembly language experience, it's mostly 64-bit MIPS with some ARM, so I'm new to Xtensa.
My goal for this is to speed up the FastLED code by using a lookup table for each nibble rather than doing a comparison for each bit. My benchmarking of my current method shows over a 25% speedup, but I want to see how far I can take it. The ee.vld.128.ip instruction looks ideal since I need to load 16 bytes of data then perform 4 word writes to the RMT memory buffer.
What am I doing wrong here? The l32i instruction accessing the exact same address works fine and loads the expected value.
I have also added code to set the wur.sar_byte and wur.accx_x registers to 0, but this makes no difference, nor do I see why this would be necessary.
Backtrace:0x403752b5:0x3fce2bc00x40375555:0x3fce2be0 0x4037558f:0x3fce2c00 0x40375a05:0x3fce2c20 0x42001486:0x3fce2c40 0x42001729:0x3fce2c60 0x42001849:0x3fce2cb0 0x4200166c:0x3fce2cf0 0x4200260d:0x3fce2d20
My code up until the crash looks like:
And the compiled code according to objdump is:
My total inline code looks like:
where pItem points to RMT memory. a14 and a15 are currently just for debugging.
Any help would be appreciated.
-Aaron
My goal for this is to speed up the FastLED code by using a lookup table for each nibble rather than doing a comparison for each bit. My benchmarking of my current method shows over a 25% speedup, but I want to see how far I can take it. The ee.vld.128.ip instruction looks ideal since I need to load 16 bytes of data then perform 4 word writes to the RMT memory buffer.
What am I doing wrong here? The l32i instruction accessing the exact same address works fine and loads the expected value.
I have also added code to set the wur.sar_byte and wur.accx_x registers to 0, but this makes no difference, nor do I see why this would be necessary.
Code: Untitled.txt Select all
entry 0x403c98d8
[ 83][D][esp32-hal-cpu.c:244] setCpuFrequencyMhz(): PLL: 480 / 6 = 80 Mhz, APB: 80000000 Hz
Guru Meditation Error: Core 1 panic'ed (LoadStoreError). Exception was unhandled.
Core 1 register dump:
PC : 0x403752b8 PS : 0x00060830 A0 : 0x80375558 A1 : 0x3fce2bc0
A2 : 0x3fc94088 A3 : 0x40378d40 A4 : 0x00000030 A5 : 0x00000000
A6 : 0x02ce3644 A7 : 0x00ffffff A8 : 0x60016800 A9 : 0x00000120
A10 : 0x00000000 A11 : 0x00000000 A12 : 0x40378d40 A13 : 0x00000000
A14 : 0x0028800a A15 : 0x40378d40 SAR : 0x00000010 EXCCAUSE: 0x00000003
EXCVADDR: 0x40378d40 LBEG : 0x400570e8 LEND : 0x400570f3 LCOUNT : 0xffffffff
My code up until the crash looks like:
Code: Untitled.asm Select all
" srli %[tmp], %[p], 4 \n"
" slli %[tmp], %[tmp], 4 \n"
" add.n %[tmp], %[tmp], %[bitTable] \n"
" mov.n a15, %[tmp] \n"
" l32i a14, %[tmp], 0 \n"
" ee.vld.128.ip q0,%[tmp],0 \n"
Code: Untitled.txt Select all
403752a2: 72b8 l32i.n a11, a2, 28
403752a4: fc6131 l32r a3, 40374428 <_iram_text_start+0x8>
403752a7: bbaa add.n a11, a11, a10
403752a9: 000bb2 l8ui a11, a11, 0
403752ac: 41c4b0 srli a12, a11, 4
403752af: 11ccc0 slli a12, a12, 4
403752b2: cc3a add.n a12, a12, a3
403752b4: 0cfd mov.n a15, a12
403752b6: 0ce8 l32i.n a14, a12, 0
403752b8: 8300c4 ee.vld.128.ip q0, a12, 0
Code: Untitled.txt Select all
register uint8_t pData = mPixelData[mCur];
register rmt_item32_t *bitTablePtr = &bitTable[0][0];
__asm__ __volatile__(
" srli %[tmp], %[p], 4 \n"
" slli %[tmp], %[tmp], 4 \n"
" add.n %[tmp], %[tmp], %[bitTable] \n"
" mov.n a15, %[tmp] \n"
" l32i a14, %[tmp], 0 \n"
" ee.vld.128.ip q0,%[tmp],0 \n"
" extui %[tmp], %[p], 0, 4 \n"
" slli %[tmp], %[tmp], 4 \n"
" add.n %[tmp], %[tmp], %[bitTable] \n"
" ee.vld.128.ip q1,%[tmp],0 \n"
" ee.movi.32.a q0, %[tmp], 3 \n"
" s32i %[tmp], %[pRmtMem], 0x0 \n"
" ee.movi.32.a q0, %[tmp], 2 \n"
" s32i %[tmp], %[pRmtMem], 0x4 \n"
" ee.movi.32.a q0, %[tmp], 1 \n"
" s32i %[tmp], %[pRmtMem], 0x8 \n"
" ee.movi.32.a q0, %[tmp], 0 \n"
" s32i %[tmp], %[pRmtMem], 0xc \n"
" ee.movi.32.a q1, %[tmp], 3 \n"
" s32i %[tmp], %[pRmtMem], 0x10 \n"
" ee.movi.32.a q1, %[tmp], 2 \n"
" s32i %[tmp], %[pRmtMem], 0x14 \n"
" ee.movi.32.a q1, %[tmp], 1 \n"
" s32i %[tmp], %[pRmtMem], 0x18 \n"
" ee.movi.32.a q1, %[tmp], 0 \n"
" s32i %[tmp], %[pRmtMem], 0x1c \n"
" addi %[pRmtMem],%[pRmtMem], 0x20 \n"
" memw \n"
: [tmp] "=&r"(tmp), [pRmtMem] "+r"(pItem)
: [bitTable] "r"(bitTablePtr), [p] "r"(pData)
: "a14", "a15");
mCur++;
Any help would be appreciated.
-Aaron