ARM glibc does not include the SDT probes that are available on other architectures. These probes are used by gdb to implement various features. You can see whether the probes are available using "readelf -n" on the .so. Be sure to examine both glibc and ld.so, as the latter has some needed probes as well. On x86-64 I get: [... other stufff ...] Notes at offset 0x001b28ac with length 0x00000294: Owner Data size Description stapsdt 0x0000003a NT_STAPSDT (SystemTap probe descriptors) Provider: libc Name: setjmp Location: 0x0000000000035a31, Base: 0x000000000017ccb0, Semaphore: 0x0000000000000000 Arguments: 8@%rdi -4@%esi 8@%rax [...] Ordinarily the fix for missing probes is simply to ensure that sys/sdt.h is available when the package is built. This file is in systemtap-sdt-devel. This should also be checked for aarch64.
ARM is not in the list of %systemtaparches and therefore is not built with system tap probes in place. Retitling as RFE.
If someone feels like trying this out and testing the probes that would be great. I expect the patch is simply: diff --git a/glibc.spec b/glibc.spec index 824f9f1..44dc593 100644 --- a/glibc.spec +++ b/glibc.spec @@ -20,7 +20,7 @@ %define biarcharches %{ix86} x86_64 ppc %{power64} s390 s390x %define debuginfocommonarches %{biarcharches} alpha alphaev6 %define multiarcharches ppc %{power64} %{ix86} x86_64 %{sparc} -%define systemtaparches %{ix86} x86_64 +%define systemtaparches %{ix86} x86_64 %{arm} # Add -s for a less verbose build output. %define silentrules PARALLELMFLAGS= However, you have to test out the resulting build.
BTW, for systemtaparches, is there a reason not to make this unconditional to all arches?
(In reply to Frank Ch. Eigler from comment #3) > BTW, for systemtaparches, is there a reason not to make this > unconditional to all arches? No reason at all. Expanding scope of issue. I'm going to unconditionally enable SDT probe usage in glibc.
I've done this in glibc-2.20.90-6.fc22