Skip to content

Fix shared_error race in validation kernels #14

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
SRO-SA wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Fix shared_error race in validation kernels #14

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 10 additions & 5 deletions cujson/parse_json_lines.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1006,22 +1006,26 @@ void validate_expand(char* pair_oc, uint32_t* index_arr, uint32_t* endIdx, int o

if( i == oc_cnt_32 - 1){
if(k+1 >= oc_cnt){
shared_error |= 1;
atomicOr(&shared_error, 1u);
// shared_error |= 1;
}else if(k+2 >= oc_cnt){
uint32_t two_chars = pair_oc[k] | pair_oc[k+1] << 8;
uint32_t shifted_two_char = two_chars << 8;
uint32_t xor_chars = (two_chars ^ shifted_two_char) & 0x0000FF00;
uint32_t error_local = __vcmpeq4(xor_chars, 0x00000600);
shared_error |= (~error_local) > 0;
atomicOr(&shared_error, (uint32_t)((~error_local) > 0));
// shared_error |= (~error_local) > 0;
endIdx[index_arr[k]] = index_arr[k+1] + lastStructuralIndex + 1;
}else if(k+3 >= oc_cnt){
shared_error |= 1;
atomicOr(&shared_error, 1u);
// shared_error |= 1;
}else{
uint32_t four_chars = pair_oc[k] | pair_oc[k+1] << 8 | pair_oc[k+2] << 16 | pair_oc[k+3] << 24;
uint32_t shifted_four_chars = four_chars << 8;
uint32_t xor_chars = (four_chars ^ shifted_four_chars) & 0xFF00FF00;
uint32_t error_local = __vcmpeq4(xor_chars, 0x06000600) & 0xFFFFFFFF;
shared_error |= (~error_local) > 0;
atomicOr(&shared_error, (uint32_t)((~error_local) > 0));
// shared_error |= (~error_local) > 0;

endIdx[index_arr[k]] = index_arr[k+1] + lastStructuralIndex + 1;
endIdx[index_arr[k+2]] = index_arr[k+3] + lastStructuralIndex + 1;
Expand All @@ -1033,7 +1037,8 @@ void validate_expand(char* pair_oc, uint32_t* index_arr, uint32_t* endIdx, int o
uint32_t xor_chars = (four_chars ^ shifted_four_chars) & 0xFF00FF00;
uint32_t error_local = __vcmpeq4(xor_chars, 0x06000600) & 0xFFFFFFFF;

shared_error |= (~error_local | order_err) > 0;
atomicOr(&shared_error, (uint32_t)((~error_local | order_err) > 0));
// shared_error |= (~error_local | order_err) > 0;
endIdx[index_arr[k]] = index_arr[k+1] + lastStructuralIndex + 1;
endIdx[index_arr[k+2]] = index_arr[k+3] + lastStructuralIndex + 1;
}
Expand Down
15 changes: 9 additions & 6 deletions cujson/parse_standard_json.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1488,7 +1488,8 @@ void validate_expand_MathAPI_new2(char* pair_oc, uint32_t* index_arr, uint32_t*
// printf("outside error: %c\n", structural_GPU[currentIndex]);
if(k+1 >= oc_cnt){
// printf("here error1: %c\n", structural_GPU[currentIndex]);
shared_error |= 1;
atomicOr(&shared_error, 1u);
// shared_error |= 1;
}else if(k+2 >= oc_cnt){
// printf("here error1: %c\n", pair_oc[k]);
// printf("here error2: %c\n", pair_oc[k+1]);
Expand All @@ -1500,14 +1501,15 @@ void validate_expand_MathAPI_new2(char* pair_oc, uint32_t* index_arr, uint32_t*
// printf("xor: %x\n", xor_chars);
uint32_t error_local = __vcmpeq4(xor_chars, 0x00000600);
// printf("err: %x\n", error_local);
shared_error |= (~error_local) > 0;
atomicOr(&shared_error, (uint32_t)((~error_local) > 0));
// shared_error |= (~error_local) > 0;

endIdx[index_arr[k]] = index_arr[k+1] + lastStructuralIndex + 1;

}else if(k+3 >= oc_cnt){
// printf("here error3: %c\n", structural_GPU[currentIndex]);
// atomicOr(&shared_error, 1);
shared_error |= 1;
atomicOr(&shared_error, 1u);
// shared_error |= 1;
}else{
// printf("here error4: %c\n", structural_GPU[currentIndex]);
// int nextIndex = index_arr[k+1];
Expand All @@ -1518,7 +1520,8 @@ void validate_expand_MathAPI_new2(char* pair_oc, uint32_t* index_arr, uint32_t*
uint32_t shifted_four_chars = four_chars << 8;
uint32_t xor_chars = (four_chars ^ shifted_four_chars) & 0xFF00FF00;
uint32_t error_local = __vcmpeq4(xor_chars, 0x06000600) & 0xFFFFFFFF;
shared_error |= (~error_local) > 0;
atomicOr(&shared_error, (uint32_t)((~error_local) > 0));
// shared_error |= (~error_local) > 0;

endIdx[index_arr[k]] = index_arr[k+1] + lastStructuralIndex + 1;
endIdx[index_arr[k+2]] = index_arr[k+3] + lastStructuralIndex + 1;
Expand All @@ -1539,7 +1542,7 @@ void validate_expand_MathAPI_new2(char* pair_oc, uint32_t* index_arr, uint32_t*
uint32_t xor_chars = (four_chars ^ shifted_four_chars) & 0xFF00FF00;
uint32_t error_local = __vcmpeq4(xor_chars, 0x06000600) & 0xFFFFFFFF;

shared_error |= (~error_local | order_err) > 0;
atomicOr(&shared_error, (uint32_t)((~error_local | order_err) > 0));

// atomicOr(&shared_error, (~error_local | order_err) > 0);

Expand Down