On Wed, Dec 15, 2010 at 5:42 PM, Howard M. Harte <hhar...@gmail.com> wrote:
> I don't know where it's crashing, tried compiling for armv5 with - > msoft-float, and still see the segfault, but no core is dumped, so no > stack trace. The device is not rooted so I can't change the ulimit to > get a core dump. Any ideas on how to get a backtrace, or even the > faulting IP? At least if I had the IP, I could look at the > disassembly. > > There should be a stacktrace in your logcat, at a minimum, which will give you the IP. If not, this either isn't a stackfault, or there is something fishy in your system image. Howe do you run the code exactly? Is it through JNI or calling a standalone executable from the adb shell? > If all else fails, I will root it (Viewsonic gTablet) and see if > changing the ulimit will produce a core dump. > > Thanks, > Howard > > On Dec 14, 2:14 am, David Turner <di...@android.com> wrote: > > First, you should try android-porting, not android-ndk > > > > Do you know what the code that crashes does exactly? You should have a > stack > > trace that tells you where the problem is located, which might give you > some > > ideas. > > What compiler flags are you using when building with the gcc-4.4.0 > > toolchain. Not using the correct one is going to produce invalid machine > > code (that's why there is an NDK in the first place). > > Also, keep in mind that Tegra2 doesn't support NEON or VFPv3-D32 (but you > > would get a SIGILL fault instead, so it's probably unrelated). > > > > On Tue, Dec 14, 2010 at 3:46 AM, Howard M. Harte <hhar...@gmail.com> > wrote: > > > > > > > > > I have some native Linux code that I built with gcc-4.4.0 from the > > > Android source repository. I build it for ARMv6, so it can run on all > > > ARM-based Android platforms, ie: ARMv6 (MSM7K, etc), ARMv7 (OMAP, > > > MSM8K, Hummingbird.) When I try to run this code on a Tegra2 CPU, I > > > get a segmentation fault. If I compile these binaries along with > > > Android for the Tegra2 (I'm using the "harmony" build in CM6.1) then > > > it runs fine. Unfortunately, these same binaries will segfault on non- > > > Tegra2 ARMv7 platforms such as OMAP. Both Tegra2 and OMAP are > > > reported as ARMv7. I don't mind having separate binaries, but I'd > > > like to understand what is going on in case there is some way to have > > > unified binaries. In any case, if Tegra2 is special, then I need to > > > find a way to differentiate it from the other ARMv7 CPUs. > > > > > I know this is not strictly related to the Android NDK, but figured > > > this would be the best place to find people who might know what the > > > issue is here. Also, I'm wondering if I might face a similar issue > > > with shared libraries that I've built using the NDK for my > > > application? > > > > > Thanks, > > > Howard > > > > > -- > > > You received this message because you are subscribed to the Google > Groups > > > "android-ndk" group. > > > To post to this group, send email to android-...@googlegroups.com. > > > To unsubscribe from this group, send email to > > > android-ndk+unsubscr...@googlegroups.com<android-ndk%2bunsubscr...@googlegroups.com> > <android-ndk%2bunsubscr...@googlegr oups.com> > > > . > > > For more options, visit this group at > > >http://groups.google.com/group/android-ndk?hl=en. > > -- > unsubscribe: > android-porting+unsubscr...@googlegroups.com<android-porting%2bunsubscr...@googlegroups.com> > website: http://groups.google.com/group/android-porting > -- unsubscribe: android-porting+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-porting
