segmentation fault - LLVM generated vector math assembly segfaulting -


the following llvm function takes 3 pointers arrays of 5 doubles , calculates c = a*b + c:

define void @my_vector_math(double* %a1, double* %b2, double* %c3) { entry:   %0 = bitcast double* %c3 <5 x double>*   %1 = load <5 x double>* %0, align 64   %2 = bitcast double* %b2 <5 x double>*   %3 = load <5 x double>* %2, align 64   %4 = bitcast double* %a1 <5 x double>*   %5 = load <5 x double>* %4, align 64   %6 = fmul <5 x double> %3, %5   %7 = fadd <5 x double> %1, %6   store <5 x double> %7, <5 x double>* %0, align 64   ret void }

it compiles down following assembly on machine (64 bit ubuntu 13.10, llvm 3.4):

  30:   c5 fb 10 42 20          vmovsd 0x20(%rdx),%xmm0   35:   c5 fb 10 4e 20          vmovsd 0x20(%rsi),%xmm1   3a:   c5 fd 28 16             vmovapd (%rsi),%ymm2   3e:   c5 fb 10 5f 20          vmovsd 0x20(%rdi),%xmm3   43:   c5 f5 59 cb             vmulpd %ymm3,%ymm1,%ymm1   47:   c5 ed 59 17             vmulpd (%rdi),%ymm2,%ymm2   4b:   c5 fd 58 c1             vaddpd %ymm1,%ymm0,%ymm0   4f:   c5 ed 58 0a             vaddpd (%rdx),%ymm2,%ymm1   53:   c5 fd 29 0a             vmovapd %ymm1,(%rdx)   57:   c5 f9 13 42 20          vmovlpd %xmm0,0x20(%rdx)   5c:   c5 f8 77                vzeroupper   5f:   c3                      retq

called arrays of size 5 segfaults, produce correct results oversized arrays:

int main() {   double a[] = {1,1,1,1,1};   double b[] = {2,2,2,2,2};   double c[] = {3,3,3,3,3};   my_vector_math(a, b, c); // segfaults   (int = 0; < 5; i++) {     printf("%f\n",c[i]);   }   return 0; }

i'm having lot of trouble figuring out why happening, pointers appreciated.

edit:

the llvm ir without optimization passes:

define void @my_vector_math(double* %a1, double* %b2, double* %c3) { entry:   %a = alloca <5 x double>   %b = alloca <5 x double>   %c = alloca <5 x double>   %c = alloca double*   %b = alloca double*   %a = alloca double*   store double* %a1, double** %a   store double* %b2, double** %b   store double* %c3, double** %c   %c4 = load double** %c   %0 = bitcast double* %c4 <5 x double>*   %1 = load <5 x double>* %0   store <5 x double> %1, <5 x double>* %c   %b5 = load double** %b   %2 = bitcast double* %b5 <5 x double>*   %3 = load <5 x double>* %2   store <5 x double> %3, <5 x double>* %b   %a6 = load double** %a   %4 = bitcast double* %a6 <5 x double>*   %5 = load <5 x double>* %4   store <5 x double> %5, <5 x double>* %a   %c7 = load <5 x double>* %c   %a8 = load <5 x double>* %a   %b9 = load <5 x double>* %b   %6 = fmul <5 x double> %a8, %b9   %7 = fadd <5 x double> %c7, %6   %c10 = load double** %c   %8 = bitcast double* %c10 <5 x double>*   store <5 x double> %7, <5 x double>* %8   ret void }

your ir claims arrays of doubles have 64-byte alignment, causes compiler generate aligned load. intended specify 8-byte alignment instead (the natural alignment of double on many platforms).


Comments

Post a Comment

Popular posts from this blog

commonjs - How to write a typescript definition file for a node module that exports a function? -

consider, toml node module can use: // toml.d.ts declare module toml { export function parse(value:string):any; } declare module "toml" { export = toml; } then: /// <reference path="../../../../../defs/toml/toml.d.ts"/> import toml = require('toml'); toml.parse(...); however, node module export single function, such 'glob' ( https://github.com/isaacs/node-glob ). the node usage of module be: var glob = require("glob") glob("*.js", {}, callback(err, files) { ... }); you'd naively expect this: // glob.d.ts declare function globs(paths:string, options:any, callback:{(err:any, files:string[]):void; ...but because typescripts 'import' semantics bit strange, seems can use 'import .. = require()' statement alias modules . trying call: /// <reference path="../../../../../defs/glob/glob.d.ts"/> import blog = require('glob'); results in: error ts2072: mod...
Read more

openid - Okta: Failed to get authorization code through API call -

i'm integrating okta own idp server using okta's api. i'm implementing authorization code flow following steps below: in own server, use /api/v1/authn endpoint sessiontoken. use sessiontoken obtain authorization calling endpoint: /oauth2/v1/authorize?client_id=" + clientid + "&sessiontoken=" + sessiontoken + "&response_type=code&response_mode=query&scope=openid&redirect_uri=" + redirecturl + "&state=evanyang&nonce=" it's supposed return response status code 302 , location header containing redirect url code value. however, keep getting response status code 200 , without location header, html body saying "you using unsupported browser." , "javascript disabled on browser." according api documentation: http://developer.okta.com/docs/api/resources/oidc.html#authentication-request , sessiontoken parameter sufficient this: an okta one-time sessiontoken. allows api-based ...
Read more

ios - Change Storyboard View using Seague -

i have following code checks connection server, , depending on want change views if no connection found. nsstring *connect = [nsstring stringwithcontentsofurl:[nsurl urlwithstring:@"http://myserver.com"] encoding:nsutf8stringencoding error:nil]; if (connect == null) { [self performseguewithidentifier:@"network_failure" sender:self]; } else { // other code here } my problem not change views. have added storyboard segue between 2 views identifier of network_failure . not have navigation controller or that, problem? thanks help! no not problem. first of all, in objectivec exists nil object oriented null . must use that. so: if(connect == nil) or better: if(!connect) so due fact nil not same of null , probably code does't enter in if scope . you have been able discover breakpoint. anyway: be sure set trigger segue (line) 2 view controller , , name network_failure (that not best name. should instead view1toview2 ) as, m...
Read more