cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
152 Views
Registered: ‎10-10-2017

snappy under sdaccel get error data

Hi, I run snappy compression/decompression (get from github.com/Xilinx/Applications) on vcu1525.
And i found that sometimes the compressed/decompression include segment of 0x00,
mostly at the end of file.while sometimes the compression/decompression result is right.
I also run the emu simulation "make check all". And the compression/decompression resultes are all right.

I check dmesg result and found messages below:
" Nov 20 08:28:33 localhost kernel: xocl:engine_status_dump: SG engine 0-C2H0-MM status: 0x00000400: R:SLAVE_ERR
Nov 20 08:28:33 localhost kernel: xocl:engine_err_handle: 0-C2H0-MM, s 0x400, aborted xfer 0xffff880459b47018, cmpl 18/180
Nov 20 08:28:33 localhost kernel: xocl:xdma_xfer_submit: xfer 0xffff880459b47018,2097152, failed, ep 0x400200000.
Nov 20 08:28:33 localhost kernel: xocl_mm_xdma mm_dma.v5.u.512: xdma_migrate_bo: DMA failed, Dumping SG Page Table"

The end of correct file is show as below: like
## "000130E0 2E 27 01 10 3B 20 60 69 6E C1 AA 00 74 65 51 49 .'..; `in...teQI
## 000130F0 23 04 6E 6F ED E8 0D BB E9 6A 24 20 4F 55 54 53 #.no.....j$ OUTS
## 00013100 49 44 45 2E 27 0E 11 18 08 20 75 6E 0E D0 13 6D IDE.'.... un...m
## 00013110 D1 25 E9 04 0D 0A DD E5 49 E2 69 F3 19 96 39 7F .%......I.i...9.
## 00013120 00 61 12 13 14 16 53 26 00 69 21 FE 04 61 20 12 .a....S&.i!..a . 0
## 0013130 00 0E 04 6F 66 41 EB 04 73 65 26 83 14 00 41 C1 ...ofA..se&...A.
## 00013140 86 0E 18 08 6D 9E 31 AA 04 27 73 12 F4 15 AD 73 ....m.1..'s....s
## 00013150 04 3F 27 16 0D 1D 08 20 61 6E 12 AC 0A 00 20 0E .?'.... an.... .
## 00013160 6B 0A 05 37 3A 22 01 04 4E 6F 16 30 0A 12 5E 21 k..7:"..No.0..^! "

The end of incorrect file like this:
## "00012FB0 71 14 21 0F 08 65 79 65 16 01 16 00 65 16 D3 31 q.!..eye....e..1
## 00012FC0 0E 4D 11 00 73 12 73 11 1D EC 01 33 2A 88 0B 2C .M..s.s....3*..,
## 00012FD0 2E 20 20 22 2D 2D 53 41 49 44 20 49 0E 38 10 38 . "--SAID I.8.8
## 00012FE0 4C 44 20 4E 4F 54 20 53 57 49 4D 2D 2D 22 20 12 LD NOT SWIM--" .
## 00012FF0 F9 27 16 17 0A C1 72 00 2C 45 B5 A1 DB E9 C2 0E .'....r.,E......
## 00013000 E8 0B 42 D5 13 12 6B 0A 2D 74 00 54 11 10 00 20 ..B...k.-t.T...
## 00013010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
## 00013020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
## 00013030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
## 00013040 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
## 00013050 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
## 00013060 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
## 00013070 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ "
and then all zeros till the end of file.

Does anyone know what is happened ?I think there is something wrong in SG dma transfer.
And i don't know how to resolve it.

My situation is Sdaccel 2018.2 ,xrt_201802.2.1.127_7.4.1708-xrt,and xilinx_vcu1525_dynamic_5_1.And OS is Centos 7.4.

0 Kudos
0 Replies