| ??? 09/24/05 05:44 Read: times |
#101477 - Yes Responding to: ???'s previous message |
Guy Lavoie said:
So technically, this can be done. Sure. Sometimes, I find that the line between a clever programming trick and a case of spaghetti code can be blurry. In most cases, it seems to be more of a question of generally accepted programming practices... My personal opinion is that there are few instances when modifying the return address is a good idea. Especially in this case where you've indicated that you want to return to a fixed address in some situations. You need to be very careful if you call the subroutine from different levels of program nesting. Consider the following sequence of events. 1. Program calls stack-modifying-subroutine. 2. Stack-modifying-subroutine pops 2 bytes off the stack (return address) and jumps to the fixed address when a carriage return is detected. 3. Program continues execution at the fixed address. But what happens in this cse: 1. Program calls a subroutine. 2. Program calls stack-modifying-subroutine. 3. Stack-modifying-subroutine pops 2 bytes off the stack (return address from step #2) and jumps to the fixed address when a carriage return is detected. 4. Program continues execution at the fixed address; but the return address from step #1 is still on the stack. So if you're not really careful and if you call the stack-modifying-subroutine from different levels, you could very easily mess up the stack and cause the stack to eventually overflow. One possibility is actually resetting the stack pointer (SP) SFR when the stack-modifying-subroutine jumps to the fixed address, but that's getting really messy and would definitely qualify as "spaghetti" in my book. This is just my opinion, though. It's not an exact science and there isn't a right or wrong answer. Be very careful if you try it. Regards, Craig Steiner |
| Topic | Author | Date |
| RET to a different address | 01/01/70 00:00 | |
| here is how pseudocode | 01/01/70 00:00 | |
| RET to a different address | 01/01/70 00:00 | |
| no flaw, but 1.000.000 gotchas | 01/01/70 00:00 | |
| That's what I wanted to know | 01/01/70 00:00 | |
| Yes | 01/01/70 00:00 | |
| "clever" | 01/01/70 00:00 | |
| OT: my wife | 01/01/70 00:00 | |
| no flaw, but seriously not recommended | 01/01/70 00:00 | |
| experience | 01/01/70 00:00 | |
| reload SP | 01/01/70 00:00 | |
| restoring stack | 01/01/70 00:00 | |
Recognisable string | 01/01/70 00:00 | |
| named return value | 01/01/70 00:00 | |
| Bad Practice | 01/01/70 00:00 | |
| Well phrased | 01/01/70 00:00 | |
| What I am doing with it | 01/01/70 00:00 | |
| try...catch | 01/01/70 00:00 | |
| setjmp / longjmp | 01/01/70 00:00 | |
| when to try ... catch | 01/01/70 00:00 | |
| the borderline | 01/01/70 00:00 | |
| Promises | 01/01/70 00:00 | |
| who cares if an exceptiom is "acceptable | 01/01/70 00:00 | |
| Parsing input data | 01/01/70 00:00 |