Wan-Teh pointed out in bug 322578 comment 57: > Looking at mozilla/security/coreconf/Linux.mk, I > think that if the cross compilation target is Linux, > we should override OS_TEST instead of CPU_ARCH. Don't > you think? On Linux, NSS only uses $(CPU_ARCH) in > CPU_TAG, which is a component of NSS's OBJDIR_NAME > and is just a name. For coreconf-based projects in general, OS_TEST seems to be the right variable to override for a same-OS-different-CPU cross compile, except for Windows, but as he pointed out to me many months ago, cross support in NSS is a little hacky anyway. At present, security/manager doesn't set OS_TEST but CPU_ARCH to drive an NSS cross build. As for Linux, OS_TEST is definitely the variable that should be overridden for cross-compilation. I suspect the same may be true for other Unixish systems, but because security/manager was overriding CPU_ARCH, some of the coreconf makefrags may have since been perverted to support that. (I just perverted the Darwin makefrag in such a way for bug 322578, but hey, there was CPU_ARCH precedent in that file!) What are the supported NSS cross build combinations? Let's figure out which targets this needs to be cleaned up for.
Doug, do you still care about cross-compiling NSS, especially with Windows CE as the target? If not, please remove yourself from the cc list.
OS: MacOS X → All
Hardware: Macintosh → All
i would love to have this happen. When Windows CE devices are more robust or there is a demand for more advanced crypto features, I would like to start using NSS again
Assignee: wtchang → nobody
QA Contact: wtchang → build
You need to log in before you can comment on or make changes to this bug.