ippiDilate crashes

flduranceau — Fri, 14 Dec 2012 16:29:59 GMT

I used the ippiDilate_8u_C1IR function for months and everything worked fine. Now, I use it in a new function and it crashes randomly but I can't get a hint on what is wrong. It crashes at MyModule.dll!_g9_Dilate_8u_C1S() +0x3bf bytes.

I use IPP 7.0 on Win7 Pro 64b.

I call ippiDilate_8u_C1IR, the image data I pass is aligned on 4bytes, mask is all filled with 1, maskSize is 3 by 3, anchor is 1, 1

Here is the Dissasembly :

0693C6F5 vpand       xmm6,xmm5,xmmword ptr [ebx+0F0h]

0693C6FD vpmaxub     xmm0,xmm0,xmm6

0693C701 je          0693CD56

0693C707 jmp _g9_Dilate_8u_C1S+17Dh (693C4EDh)

0693C70C mov         dword ptr [esp+0Ch],ecx

0693C710 mov         ecx,dword ptr [esp+50h]

0693C714 mov         edi,ebx

0693C716 mov         edx,dword ptr [esp+58h]

0693C71A vpxor       xmm0,xmm0,xmm0

0693C71E mov         dword ptr [esp],eax

0693C721 mov         dword ptr [esp+8],esi

0693C725 mov         dword ptr [esp+20h],ebx

0693C729 mov         esi,dword ptr [esp+54h]

0693C72D mov         ebx,ecx

crashes here ----> 0693C72F vmovdqu     xmm3,xmmword ptr [edi]

0693C733 lea         eax,[edi+10h]

0693C736 jmp         _g9_Dilate_8u_C1S+3D5h (693C745h)

0693C738 vmovdqa     xmm3,xmm1

0693C73C add         eax,10h

0693C73F add         ebx,100h

0693C745 vpand       xmm4,xmm3,xmmword ptr [ebx]

0693C749 cmp         esi,1

0693C74C vmovdqu     xmm1,xmmword ptr [eax]

0693C750 vpmaxub     xmm0,xmm0,xmm4

0693C754 je          0693CD8A

0693C75A vpsrldq     xmm3,xmm3,1

0693C75F cmp         esi,2

0693C762 vpslldq     xmm4,xmm1,0Fh

0693C767 vpor        xmm3,xmm3,xmm4

0693C76B vpand       xmm5,xmm3,xmmword ptr [ebx+10h]

And here is the content of my registers :

EAX = 0000003C EBX = 0324A6F0 ECX = 0324A6F0 EDX = 00000003 ESI = 00000003 EDI = 1F15FC5F EIP = 0693C72F ESP = 0324A68C EBP = 1F15FC5F EFL = 00010206

CS = 0023 DS = 002B ES = 002B SS = 002B FS = 0053 GS = 002B

ST0 = +0.0000000000000000e+0000   ST1 = +0.0000000000000000e+0000   ST2 = +0.0000000000000000e+0000   ST3 = +0.0000000000000000e+0000   ST4 = +0.0000000000000000e+0000   ST5 = +7.9000000000000000e+0001   ST6 = +2.8600000000000000e+0002   ST7 = +0.0000000000000000e+0000   CTRL = 027F STAT = 0120 TAGS = FFFF EIP = 067410AC EDO = 0324AAA4

MM0 = 0000000000000000 MM1 = 0000000000000000 MM2 = 0000000000000000 MM3 = 0000000000000000 MM4 = 0000000000000000 MM5 = 9E00000000000000 MM6 = 8F00000000000000 MM7 = 0000000000000000

XMM0 = 00000000000000000000000000000000 XMM1 = 00000000000000000000000000000000 XMM2 = 00000000000000000000000000000000 XMM3 = 00000000000000000000000000000000 XMM4 = 00000000000000000000000000000000 XMM5 = 00000000000000000000000000000000 XMM6 = FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF XMM7 = 00000000000000000000000000000000 XMM00 = +0.00000E+000      XMM01 = +0.00000E+000      XMM02 = +0.00000E+000      XMM03 = +0.00000E+000      XMM10 = +0.00000E+000      XMM11 = +0.00000E+000      XMM12 = +0.00000E+000      XMM13 = +0.00000E+000      XMM20 = +0.00000E+000      XMM21 = +0.00000E+000      XMM22 = +0.00000E+000      XMM23 = +0.00000E+000      XMM30 = +0.00000E+000      XMM31 = +0.00000E+000      XMM32 = +0.00000E+000      XMM33 = +0.00000E+000      XMM40 = +0.00000E+000      XMM41 = +0.00000E+000      XMM42 = +0.00000E+000      XMM43 = +0.00000E+000      XMM50 = +0.00000E+000      XMM51 = +0.00000E+000      XMM52 = +0.00000E+000      XMM53 = +0.00000E+000      XMM60 = -1.#QNANE+000      XMM61 = -1.#QNANE+000      XMM62 = -1.#QNANE+000      XMM63 = -1.#QNANE+000      XMM70 = +0.00000E+000      XMM71 = +0.00000E+000      XMM72 = +0.00000E+000      XMM73 = +0.00000E+000      MXCSR = 00001F80

XMM0DL = +0.0000000000000000E+000         XMM0DH = +0.0000000000000000E+000         XMM1DL = +0.0000000000000000E+000         XMM1DH = +0.0000000000000000E+000         XMM2DL = +0.0000000000000000E+000         XMM2DH = +0.0000000000000000E+000         XMM3DL = +0.0000000000000000E+000         XMM3DH = +0.0000000000000000E+000         XMM4DL = +0.0000000000000000E+000         XMM4DH = +0.0000000000000000E+000         XMM5DL = +0.0000000000000000E+000         XMM5DH = +0.0000000000000000E+000         XMM6DL = -1.#QNAN00000000000E+000         XMM6DH = -1.#QNAN00000000000E+000         XMM7DL = +0.0000000000000000E+000         XMM7DH = +0.0000000000000000E+000

YMM0 = F1F2000000000000654C00000000000000000000000000000000000000000000 YMM1 = 654C000000000000000000000000000000000000000000000000000000000000 YMM2 = 0000000000000000F1F200000000000000000000000000000000000000000000 YMM3 = 000A0000000000006E6500000000000000000000000000000000000000000000 YMM4 = 45790000000000006F7200000000000000000000000000000000000000000000 YMM5 = 00010000000000006E6F00000000000000000000000000000000000000000000 YMM6 = 00010000000000006174000000000000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF YMM7 = 00010000000000006B7200000000000000000000000000000000000000000000 YMM0I0 = 0000000000000000 YMM0I1 = 0000000000000000 YMM0I2 = 654C000000000000 YMM0I3 = F1F2000000000000 YMM1I0 = 0000000000000000 YMM1I1 = 0000000000000000 YMM1I2 = 0000000000000000 YMM1I3 = 654C000000000000 YMM2I0 = 0000000000000000 YMM2I1 = 0000000000000000 YMM2I2 = F1F2000000000000 YMM2I3 = 0000000000000000 YMM3I0 = 0000000000000000 YMM3I1 = 0000000000000000 YMM3I2 = 6E65000000000000 YMM3I3 = 000A000000000000 YMM4I0 = 0000000000000000 YMM4I1 = 0000000000000000 YMM4I2 = 6F72000000000000 YMM4I3 = 4579000000000000 YMM5I0 = 0000000000000000 YMM5I1 = 0000000000000000 YMM5I2 = 6E6F000000000000 YMM5I3 = 0001000000000000

YMM6I0 = FFFFFFFFFFFFFFFF YMM6I1 = FFFFFFFFFFFFFFFF YMM6I2 = 6174000000000000 YMM6I3 = 0001000000000000 YMM7I0 = 0000000000000000 YMM7I1 = 0000000000000000 YMM7I2 = 6B72000000000000 YMM7I3 = 0001000000000000

YMM0F0 = +0.00000E+000      YMM0F1 = +0.00000E+000      YMM0F2 = +0.00000E+000      YMM0F3 = +0.00000E+000      YMM0F4 = +0.00000E+000      YMM0F5 = +6.02102E+022      YMM0F6 = +0.00000E+000      YMM0F7 = -2.39665E+030      YMM1F0 = +0.00000E+000      YMM1F1 = +0.00000E+000      YMM1F2 = +0.00000E+000      YMM1F3 = +0.00000E+000      YMM1F4 = +0.00000E+000      YMM1F5 = +0.00000E+000      YMM1F6 = +0.00000E+000      YMM1F7 = +6.02102E+022      YMM2F0 = +0.00000E+000      YMM2F1 = +0.00000E+000      YMM2F2 = +0.00000E+000      YMM2F3 = +0.00000E+000      YMM2F4 = +0.00000E+000      YMM2F5 = -2.39665E+030      YMM2F6 = +0.00000E+000      YMM2F7 = +0.00000E+000      YMM3F0 = +0.00000E+000      YMM3F1 = +0.00000E+000      YMM3F2 = +0.00000E+000      YMM3F3 = +0.00000E+000      YMM3F4 = +0.00000E+000      YMM3F5 = +1.77180E+028      YMM3F6 = +0.00000E+000      YMM3F7 = +9.2E-040#DEN      YMM4F0 = +0.00000E+000      YMM4F1 = +0.00000E+000      YMM4F2 = +0.00000E+000      YMM4F3 = +0.00000E+000      YMM4F4 = +0.00000E+000      YMM4F5 = +7.48954E+028      YMM4F6 = +0.00000E+000      YMM4F7 = +3.98400E+003      YMM5F0 = +0.00000E+000      YMM5F1 = +0.00000E+000

YMM5F2 = +0.00000E+000      YMM5F3 = +0.00000E+000      YMM5F4 = +0.00000E+000      YMM5F5 = +1.84917E+028      YMM5F6 = +0.00000E+000      YMM5F7 = +9.2E-041#DEN      YMM6F0 = -1.#QNANE+000      YMM6F1 = -1.#QNANE+000      YMM6F2 = -1.#QNANE+000      YMM6F3 = -1.#QNANE+000      YMM6F4 = +0.00000E+000      YMM6F5 = +2.81313E+020      YMM6F6 = +0.00000E+000      YMM6F7 = +9.2E-041#DEN      YMM7F0 = +0.00000E+000      YMM7F1 = +0.00000E+000      YMM7F2 = +0.00000E+000      YMM7F3 = +0.00000E+000      YMM7F4 = +0.00000E+000      YMM7F5 = +2.92560E+026      YMM7F6 = +0.00000E+000      YMM7F7 = +9.2E-041#DEN      YMM0D0 = +0.0000000000000000E+000         YMM0D1 = +0.0000000000000000E+000         YMM0D2 = +9.0770653069271721E+179         YMM0D3 = -7.5015162369898361E+240         YMM1D0 = +0.0000000000000000E+000         YMM1D1 = +0.0000000000000000E+000         YMM1D2 = +0.0000000000000000E+000         YMM1D3 = +9.0770653069271721E+179         YMM2D0 = +0.0000000000000000E+000         YMM2D1 = +0.0000000000000000E+000         YMM2D2 = -7.5015162369898361E+240         YMM2D3 = +0.0000000000000000E+000         YMM3D0 = +0.0000000000000000E+000

YMM3D1 = +0.0000000000000000E+000         YMM3D2 = +6.0727596168061509E+223         YMM3D3 = +1.390671161567E-308#DEN         YMM4D0 = +0.0000000000000000E+000         YMM4D1 = +0.0000000000000000E+000         YMM4D2 = +6.8225892728058498E+228         YMM4D3 = +4.8357032784585167E+026         YMM5D0 = +0.0000000000000000E+000         YMM5D1 = +0.0000000000000000E+000         YMM5D2 = +8.9645499105233656E+223         YMM5D3 = +1.390671161567E-309#DEN         YMM6D0 = -1.#QNAN00000000000E+000         YMM6D1 = -1.#QNAN00000000000E+000         YMM6D2 = +2.8118211215894977E+161         YMM6D3 = +1.390671161567E-309#DEN         YMM7D0 = +0.0000000000000000E+000         YMM7D1 = +0.0000000000000000E+000         YMM7D2 = +3.6985330557762001E+209         YMM7D3 = +1.390671161567E-309#DEN

OV = 0 UP = 0 EI = 1 PL = 0 ZR = 0 AC = 0 PE = 1 CY = 0

1F15FC5F = ????????????????????????????????

　

Thanks for any help or clue to find out what is wrong in my code.

Francis

Hi Francis,

SergeyKostrov — Fri, 14 Dec 2012 20:11:00 GMT

Hi Francis, >>... It crashes at MyModule.dll!_g9_Dilate_8u_C1S() +0x3bf bytes... Unfortunately, it is not enough to understand what is wrong. >>...Thanks for any help or clue to find out what is wrong in my code... Any chance to post a reproducer that simulates your real processing and crashes? Best regards, Sergey

Hi!

Bernard — Fri, 14 Dec 2012 21:55:00 GMT

Hi! While doing single step debugging please post the content of the memory pointed by this instructions :vpand xmm6,xmm5,xmmword ptr [ebx+0F0h] What is the memory content of "ebx+0F0h? Please observe the different values(stack addresses) which are pointed by ebx register and try to locate their content in the memory window. What type of crash are you experiencing? Is this so called memory access violation code 0xc0000005 or something else? Also crucial information needed to understand the culprit is related to this mov instruction : mov edi,ebx

Finally, we have figured out

flduranceau — Mon, 17 Dec 2012 15:03:00 GMT

Finally, we have figured out that when we deactivate another of our threads, we are fine. So it is not related at all with the dilate and it's really a problem on our side of a thread that is writing over our memory. iliyapolak, we were getting an access violation code 0xc0000005 and looking closer to the registers, I think that our anchor parameter was getting override by some junk from our bugged thread. Thanks for the support. I think we can remove this whole topic since it's not relevant anymore. Francis

You could update the title of

SergeyKostrov — Mon, 17 Dec 2012 15:41:42 GMT

You could update the title of the thread to: '[SOLVED] ippiDilate crashes' or '[FIXED] ippiDilate crashes'. Thanks in advance.

>>> think that our anchor

Bernard — Mon, 17 Dec 2012 19:03:57 GMT

>>> think that our anchor parameter was getting override by some junk from our bugged thread.>>> Yes it was quite possible.Probably leftover from the other thread's called functions, or improper functions cleanup.

topic >>> think that our anchor in Intel® Integrated Performance Primitives

ippiDilate crashes

Hi Francis,

Hi!

Finally, we have figured out

You could update the title of

>>> think that our anchor