pghs975uc
/
infer_clone
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from 'b1698ab0ea'
${ noResults }
infer_clone/infer/tests/codetoanalyze/c/frontend/switchstmt/switch_unroll.c.dot

109 lines
8.3 KiB
Raw Normal View History Unescape

[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
/* @generated */
digraph cfg {
Get rid of Declare_locals Reviewed By: jeremydubreil Differential Revision: D9234331 fbshipit-source-id: 70443eabe
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_1" [label="1: Start unroll_loop\nFormals: n:int\nLocals: loop:int ret:int \n " color=yellow style=filled]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_1" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_25" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_2" [label="2: Exit unroll_loop \n " color=yellow style=filled]
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_3" [label="3: Return Stmt \n n$0=*&ret:int [line 24, column 10]\n *&return:int=n$0 [line 24, column 3]\n NULLIFY(&ret); [line 24, column 3]\n EXIT_SCOPE(n$0,ret); [line 24, column 3]\n APPLY_ABSTRACTION; [line 24, column 3]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_3" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_2" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_4" [label="4: SwitchStmt \n n$1=*&n:int [line 10, column 11]\n NULLIFY(&n); [line 10, column 11]\n EXIT_SCOPE(n); [line 10, column 11]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_4" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_22" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_4" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_23" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_5" [label="5: + \n " ]
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_5" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_15" ;
Change license to MIT Summary: Change the license of the source code from BSD + PATENTS to MIT. Change `checkCopyright` to reflect the new license and learn some new file types. Generated with: ``` git grep BSD | xargs -n 1 ./scripts/checkCopyright -i ``` Reviewed By: jeremydubreil, mbouaziz, jberdine Differential Revision: D8071249 fbshipit-source-id: 97ca23a
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_6" [label="6: BinaryOperatorStmt: GT \n n$2=*&loop:int [line 22, column 16]\n *&loop:int=(n$2 - 1) [line 22, column 16]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_6" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_7" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_6" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_8" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_7" [label="7: Prune (true branch, do while) \n PRUNE(((n$2 - 1) > 0), true); [line 22, column 16]\n EXIT_SCOPE(n$2); [line 22, column 16]\n APPLY_ABSTRACTION; [line 22, column 16]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_7" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_5" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_8" [label="8: Prune (false branch, do while) \n PRUNE(!((n$2 - 1) > 0), false); [line 22, column 16]\n NULLIFY(&loop); [line 22, column 16]\n EXIT_SCOPE(n$2,loop); [line 22, column 16]\n APPLY_ABSTRACTION; [line 22, column 16]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_8" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_3" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_9" [label="9: UnaryOperator \n n$3=*&ret:int [line 21, column 11]\n *&ret:int=(n$3 + 1) [line 21, column 11]\n EXIT_SCOPE(n$3); [line 21, column 11]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_9" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_6" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_10" [label="10: + \n " ]
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_10" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_9" ;
[cfg] print dotty *after* pre-analysis Summary: Useful to understand the changes in the pre-analysis, or to inspect the CFG that checkers actually get. This means that the pre-analysis always runs when we output the dotty, but I don't really see a reason why not. In fact, we could probably *always* store the CFGs as pre-analysed. Reviewed By: mbouaziz Differential Revision: D13102952 fbshipit-source-id: 89f3102ec
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_11" [label="11: Prune (true branch, if) \n PRUNE(1, true); [line 16, column 15]\n APPLY_ABSTRACTION; [line 16, column 15]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_11" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_13" ;
[cfg] print dotty *after* pre-analysis Summary: Useful to understand the changes in the pre-analysis, or to inspect the CFG that checkers actually get. This means that the pre-analysis always runs when we output the dotty, but I don't really see a reason why not. In fact, we could probably *always* store the CFGs as pre-analysed. Reviewed By: mbouaziz Differential Revision: D13102952 fbshipit-source-id: 89f3102ec
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_12" [label="12: Prune (false branch, if) \n PRUNE(!1, false); [line 16, column 15]\n APPLY_ABSTRACTION; [line 16, column 15]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_12" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_10" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_13" [label="13: UnaryOperator \n n$4=*&ret:int [line 18, column 15]\n *&ret:int=(n$4 + 1) [line 18, column 15]\n EXIT_SCOPE(n$4); [line 18, column 15]\n APPLY_ABSTRACTION; [line 18, column 15]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_13" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_10" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_14" [label="14: UnaryOperator \n n$7=*&ret:int [line 15, column 11]\n *&ret:int=(n$7 + 1) [line 15, column 11]\n EXIT_SCOPE(n$7); [line 15, column 11]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_14" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_11" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_14" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_12" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_15" [label="15: UnaryOperator \n n$8=*&ret:int [line 13, column 9]\n *&ret:int=(n$8 + 1) [line 13, column 9]\n EXIT_SCOPE(n$8); [line 13, column 9]\n APPLY_ABSTRACTION; [line 13, column 9]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_15" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_14" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_16" [label="16: Prune (true branch, switch) \n PRUNE(((n$1 % 8) == 1), true); [line 20, column 9]\n EXIT_SCOPE(n$1); [line 20, column 9]\n APPLY_ABSTRACTION; [line 20, column 9]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_16" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_9" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_17" [label="17: Prune (false branch, switch) \n PRUNE(!((n$1 % 8) == 1), false); [line 20, column 9]\n NULLIFY(&loop); [line 20, column 9]\n EXIT_SCOPE(n$1,loop); [line 20, column 9]\n APPLY_ABSTRACTION; [line 20, column 9]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_17" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_3" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_18" [label="18: Prune (true branch, switch) \n PRUNE(((n$1 % 8) == 2), true); [line 17, column 13]\n EXIT_SCOPE(n$1); [line 17, column 13]\n APPLY_ABSTRACTION; [line 17, column 13]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_18" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_13" ;
Change license to MIT Summary: Change the license of the source code from BSD + PATENTS to MIT. Change `checkCopyright` to reflect the new license and learn some new file types. Generated with: ``` git grep BSD | xargs -n 1 ./scripts/checkCopyright -i ``` Reviewed By: jeremydubreil, mbouaziz, jberdine Differential Revision: D8071249 fbshipit-source-id: 97ca23a
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_19" [label="19: Prune (false branch, switch) \n PRUNE(!((n$1 % 8) == 2), false); [line 17, column 13]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_19" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_16" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_19" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_17" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_20" [label="20: Prune (true branch, switch) \n PRUNE(((n$1 % 8) == 3), true); [line 14, column 9]\n EXIT_SCOPE(n$1); [line 14, column 9]\n APPLY_ABSTRACTION; [line 14, column 9]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_20" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_14" ;
Change license to MIT Summary: Change the license of the source code from BSD + PATENTS to MIT. Change `checkCopyright` to reflect the new license and learn some new file types. Generated with: ``` git grep BSD | xargs -n 1 ./scripts/checkCopyright -i ``` Reviewed By: jeremydubreil, mbouaziz, jberdine Differential Revision: D8071249 fbshipit-source-id: 97ca23a
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_21" [label="21: Prune (false branch, switch) \n PRUNE(!((n$1 % 8) == 3), false); [line 14, column 9]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_21" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_18" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_21" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_19" ;
[preanalysis] do not delay killing variables taken by reference Summary: Before, the liveness pre-analysis would place extra instructions in the CFG for either: 1. marking an `Ident.t` as dead, or 2. marking a `Pvar.t` as `= 0` But we have no way of marking pvars dead without setting them to 0. This is bad because setting pvars to 0 is not possible everywhere they are dead. Indeed, we only do it when we haven't seen their address being taken anyway. This prevents the following situation, recorded in our tests: ``` int address_taken() { int** x; int* y; int i = 7; y = &i; x = &y; // if we don't reason about taken addresses while adding nullify instructions, // we'll add // `nullify(y)` here and report a false NPE on the next line return **x; } ``` So we want to mark pvars as dead without nullifying them. This diff extends the `Remove_temps` SIL instruction to accept pvars as well, and so renames it to `ExitScope`. Reviewed By: da319 Differential Revision: D13102953 fbshipit-source-id: aa7f03a52
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_22" [label="22: Prune (true branch, switch) \n PRUNE(((n$1 % 8) == 0), true); [line 11, column 5]\n EXIT_SCOPE(n$1); [line 11, column 5]\n APPLY_ABSTRACTION; [line 11, column 5]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_22" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_5" ;
Change license to MIT Summary: Change the license of the source code from BSD + PATENTS to MIT. Change `checkCopyright` to reflect the new license and learn some new file types. Generated with: ``` git grep BSD | xargs -n 1 ./scripts/checkCopyright -i ``` Reviewed By: jeremydubreil, mbouaziz, jberdine Differential Revision: D8071249 fbshipit-source-id: 97ca23a
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_23" [label="23: Prune (false branch, switch) \n PRUNE(!((n$1 % 8) == 0), false); [line 11, column 5]\n " shape="invhouse"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_23" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_20" ;
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_23" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_21" ;
[SIL] change `variable_initialization()` builtin to a new auxiliary instruction Summary: Instead of emitting an ad-hoc builtin on variable declaration emit a new metadata instruction. This allows us to remove the code matching on that ad-hoc builtin that had to be inserted in several checkers. Inferbo & pulse used that information meaningfully and had to undergo some minor changes to cope with the new metada instruction. Reviewed By: ezgicicek Differential Revision: D14833100 fbshipit-source-id: 9b3009d22
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_24" [label="24: DeclStmt \n VARIABLE_DECLARED(loop:int); [line 9, column 3]\n n$11=*&n:int [line 9, column 14]\n *&loop:int=(n$11 + (3 / 4)) [line 9, column 3]\n EXIT_SCOPE(n$11); [line 9, column 3]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_24" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_4" ;
[SIL] change `variable_initialization()` builtin to a new auxiliary instruction Summary: Instead of emitting an ad-hoc builtin on variable declaration emit a new metadata instruction. This allows us to remove the code matching on that ad-hoc builtin that had to be inserted in several checkers. Inferbo & pulse used that information meaningfully and had to undergo some minor changes to cope with the new metada instruction. Reviewed By: ezgicicek Differential Revision: D14833100 fbshipit-source-id: 9b3009d22
6 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_25" [label="25: DeclStmt \n VARIABLE_DECLARED(ret:int); [line 8, column 3]\n *&ret:int=0 [line 8, column 3]\n " shape="box"]
[clang] make switch statement translation more robust Summary: Labels inside switch statements were causing havoc (see test), and the translation of switch statements in general could be improved to handle more cases. It turns out that `case` (and `default`) statements are more or less fancy labels into the code. In other words, if you erase all the `case XXX:` and `default:` strings in the `switch` statement you get the real structure of the program, and `switch` just jumps straight to the first `case` directives (and to the second if the first one is not satisfied, etc. until all `case`/`default` have been considered). This suggests an alternative implementation: translate the body of the `switch` and simply record the list of switch cases inside that body, along with where they point to. Then post-process this list to construct the control flow of the `switch`, which points into the control-flow of the `body`. In order not to modify every function in `CTrans` to propagate the current list of cases, I created an ugly `ref` inside `SwitchCase` instead (but it cannot be directly accessed and it's guaranteed to be well-parenthesised wrt nested switches by the `SwitchCase` API so it's not too bad). [unrelated] Also make translation failures output more information about what exactly in the source code is causing the crash, and the ancestors in the AST that lead to the crash site. Reviewed By: martinoluca Differential Revision: D8011046 fbshipit-source-id: 8455090
7 years ago
"unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_25" -> "unroll_loop.7d9e50ecf5e5106a8dd5deee005639d6_24" ;
}