Test suite failure: Strip -s libdwarf1c.so

[pipcet]

This one is weird. In the nm -D output,

f0000000 R _etext

changes to

f0000000 A _etext

during the strip -s.

[pipcet]

Apparently a linker script issue: strip -s doesn't touch the dynamic symbol table, but it strips sections; if a symbol, such as _etext, lives in a section whose index becomes invalid during the stripping, it's not adjusted.

I'm not sure whether this is really intentional behavior of strip or a bug.