|
From: Naveen K. <g_n...@ya...> - 2005-09-05 15:24:17
|
void *mmap(void *addr, size_t len, int prot, int flags, int fildes, off_t off); One of the flags that can be passed to mmap on solaris is MAP_ALIGN. If that flag is set then the addr arg specifies the alignment of the return address of mmap. How easy can this functionality be hacked into VG_(find_map_space) ? Will this affect anything else ? Thanks Naveen ______________________________________________________ Click here to donate to the Hurricane Katrina relief effort. http://store.yahoo.com/redcross-donate3/ |
|
From: Nicholas N. <nj...@cs...> - 2005-09-13 19:59:23
|
On Mon, 5 Sep 2005, Naveen Kumar wrote: > void *mmap(void *addr, size_t len, int prot, int > flags, int fildes, off_t off); > > One of the flags that can be passed to mmap on solaris > is MAP_ALIGN. If that flag is set then the addr arg > specifies the alignment of the return address of mmap. > How easy can this functionality be hacked into > VG_(find_map_space) ? Will this affect anything else ? It's probably not that hard, it would just be a few lines in VG_(find_map_space)(), no? Note that Julian's work on the ASPACEM branch will affect this. The changes will hopefully make porting to other OSes easier (Greg Parker has confirmed that this is so for Darwin) but it means you will have to pay attention to two branches, ASPACEM and the trunk. Nick |