[PATCH 2/3] gdb: Don't reorder line table entries too much when sorting.

[Tom de Vries]

On 24-01-2020 18:35, Tom Tromey wrote:
>>>>>> "Andrew" == Andrew Burgess <andrew.burgess@embecosm.com> writes:
> 
> Andrew> Don't reorder line table entries for the same address when sorting the
> Andrew> line table, maintain the compiler given line order.  Usually this will
> Andrew> reflect the order in which lines are conceptually encountered at a
> Andrew> given address.
> 
> Thanks for the long explanation and the patch.
> 
> I had a couple minor nits.
> 
> Andrew> -	  /* Like the pending blocks, the line table may be
> Andrew> -	     scrambled in reordered executables.  Sort it if
> Andrew> -	     OBJF_REORDERED is true.  */
> Andrew> +	  const auto lte_is_less_than
> Andrew> +	    = [] (const linetable_entry &ln1,
> Andrew> +		  const linetable_entry &ln2)->bool
> 
> I'd put spaces around the "->".
> 
> 
> Andrew> +	      {
> Andrew> +		/* Note: this code does not assume that CORE_ADDRs can fit
> Andrew> +		   in ints.  Please keep it that way.  */
> Andrew> +		return (ln1.pc < ln2.pc);
> 
> I don't think this comment adds anything any more.  IMO it can just be
> dropped.

This commit caused the following regressions with target board readnow:
...
+FAIL: gdb.ada/bp_c_mixed_case.exp: break <NoDebugMixedCaseFunc>
+FAIL: gdb.arch/amd64-invalid-stack-top.exp: first backtrace, with error
message
+FAIL: gdb.arch/amd64-invalid-stack-top.exp: second backtrace, with
error message
+FAIL: gdb.arch/amd64-invalid-stack-top.exp: check mi -stack-list-frames
command, first time
+FAIL: gdb.arch/amd64-invalid-stack-top.exp: check mi -stack-list-frames
command, second time
+FAIL: gdb.arch/i386-bp_permanent.exp: single-step past permanent breakpoint
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=0:
final_debug=0: step
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=0:
final_debug=0: set-break: after resolving: break final
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=0:
final_debug=0: set-break: after resolving: break gnu_ifunc
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=0:
final_debug=0: set-break: after resolving: info breakpoints
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=1:
final_debug=0: step
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=1:
final_debug=0: set-break: after resolving: break final
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=1:
final_debug=0: set-break: after resolving: break gnu_ifunc
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=1:
final_debug=0: set-break: after resolving: info breakpoints
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=0:
final_debug=0: step
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=0:
final_debug=0: set-break: after resolving: break final
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=0:
final_debug=0: set-break: after resolving: break gnu_ifunc
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=0:
final_debug=0: set-break: after resolving: info breakpoints
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=1:
final_debug=0: step
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=1:
final_debug=0: set-break: after resolving: break final
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=1:
final_debug=0: set-break: after resolving: break gnu_ifunc
+FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=1:
final_debug=0: set-break: after resolving: info breakpoints
+FAIL: gdb.base/solib-weak.exp: run to breakpoint - lib1 nodebug, lib2
nodebug, lib1 first
+FAIL: gdb.base/solib-weak.exp: run to breakpoint - lib1 nodebug, lib2
nodebug, lib2 first
+FAIL: gdb.base/solib-weak.exp: run to breakpoint - lib1 nodebug, lib2
debug, lib1 first
+FAIL: gdb.base/solib-weak.exp: run to breakpoint - lib1 debug, lib2
nodebug, lib2 first
+FAIL: gdb.base/step-symless.exp: step
+FAIL: gdb.base/until-nodebug.exp: until 1
+FAIL: gdb.base/until-nodebug.exp: until 2 (the program exited)
+FAIL: gdb.btrace/unknown_functions.exp: flat
+FAIL: gdb.btrace/unknown_functions.exp: indented
+FAIL: gdb.cp/minsym-fallback.exp: break C::f()
+FAIL: gdb.cp/minsym-fallback.exp: break C::operator()()
+FAIL: gdb.reverse/singlejmp-reverse.exp: reverse-step
+FAIL: gdb.reverse/singlejmp-reverse.exp: reverse-next
...

As well as the following progressions:
...
+PASS: gdb.gdb/complaints.exp: breakpoint in captured_command_loop
+PASS: gdb.gdb/python-interrupts.exp: breakpoint in captured_command_loop
+PASS: gdb.gdb/python-selftest.exp: breakpoint in captured_command_loop
+PASS: gdb.gdb/selftest.exp: breakpoint in captured_main
...

The first regression is noted in PR25858 - "[readnow] FAIL:
gdb.ada/bp_c_mixed_case.exp: break <NoDebugMixedCaseFunc>" (
https://sourceware.org/bugzilla/show_bug.cgi?id=25858 ).

Thanks,
- Tom