This shows you the differences between two versions of the page.
codebase:remove_m32 [2018/08/23 07:29] vodur created |
codebase:remove_m32 [2018/08/23 07:30] (current) vodur |
||
---|---|---|---|
Line 1: | Line 1: | ||
- | Sender Vodur~\\ | + | <code text> |
- | Date Fri Jun 16 00:49:37 2017~\\ | + | Sender Vodur~ |
- | Stamp 1497588608\\ | + | Date Fri Jun 16 00:49:37 2017~ |
- | Expire 1507956584\\ | + | Stamp 1497588608 |
- | To imm~\\ | + | Expire 1507956584 |
- | Subject removing -m32 flag~\\ | + | To imm~ |
- | Text\\ | + | Subject removing -m32 flag~ |
- | I propose to remove -m32 flag from makefile.\\ | + | Text |
- | What we gain from this is mostly future pain of setting up 32-bit build\\ | + | I propose to remove -m32 flag from makefile. |
- | dependencies on 64-bit systems (as we've all done by now).\\ | + | What we gain from this is mostly future pain of setting up 32-bit build |
- | Potential issues, as far as I understand, would be if we have code that assumes\\ | + | dependencies on 64-bit systems (as we've all done by now). |
+ | Potential issues, as far as I understand, would be if we have code that assumes | ||
size of types. Here are the sizes with and without -m32. | size of types. Here are the sizes with and without -m32. | ||
- | -m32 no -m32\\ | + | -m32 no -m32 |
- | The size of int 4 4\\ | + | The size of int 4 4 |
- | The size of short 2 2\\ | + | The size of short 2 2 |
- | The size of long 4 8\\ | + | The size of long 4 8 |
- | The size of float 4 4\\ | + | The size of float 4 4 |
- | The size of double 8 8\\ | + | The size of double 8 8 |
- | The size of long double 12 16\\ | + | The size of long double 12 16 |
- | The size of char 1 1\\ | + | The size of char 1 1 |
- | The size of enum 4 4\\ | + | The size of enum 4 4 |
- | The size of pointer 4 8\\ | + | The size of pointer 4 8 |
- | So only long, long double, and pointer sizes have changed. This being the case\\ | + | So only long, long double, and pointer sizes have changed. This being the case |
I say the risk of issues from size change is low. | I say the risk of issues from size change is low. | ||
- | I verified code can compile and run just fine without -m32. Only issue I see at\\ | + | I verified code can compile and run just fine without -m32. Only issue I see at |
- | the moment is there is casting between void * and int in hunt.c which gets a\\ | + | the moment is there is casting between void * and int in hunt.c which gets a |
warning since they are different sizes with no -m32. | warning since they are different sizes with no -m32. | ||
Thoughts? | Thoughts? | ||
- | P.S.: On a related note, we are currently using gcc 6.2.1 on the host right now\\ | + | P.S.: On a related note, we are currently using gcc 6.2.1 on the host right now |
- | (through scl devtoolset-6), which for some reason can't link libstdc++ 32-bit,\\ | + | (through scl devtoolset-6), which for some reason can't link libstdc++ 32-bit, |
- | so if/when we do start adding C++ code, we would need to turn -m32 off or go\\ | + | so if/when we do start adding C++ code, we would need to turn -m32 off or go |
- | back to gcc 4.1.2.\\ | + | back to gcc 4.1.2. |
~ | ~ | ||
+ | </code> | ||