Re: "set record instruction-history" ?

[Pedro Alves <palves@redhat.com>]

On 03/26/2013 07:46 AM, Metzger, Markus T wrote:

> Shouldn't we map 0 to INT_MAX before committing the new value?

That's already done for us before this function is reached, by the
set/show machinery.  "commit" here refers to copying the set/show
command's control variable value to the real setting variable
(record_(call|insn)_history_size).

> The functionality should be covered by the gdb.btrace test suite.  Do you want
> to commit your patch or do you want me to test and then commit it?

Could you test and report back please?  I believe my laptop's cpu is
one of those where branch tracing is borked and ended up disabled.

...
(gdb) record btrace
Target does not support branch tracing.
(gdb) testcase ../../../src/gdb/testsuite/gdb.btrace/enable.exp completed in 1 seconds
Running ../../../src/gdb/testsuite/gdb.btrace/function_call_history.exp ...
testcase ../../../src/gdb/testsuite/gdb.btrace/function_call_history.exp completed in 0 seconds
Running ../../../src/gdb/testsuite/gdb.btrace/instruction_history.exp ...
testcase ../../../src/gdb/testsuite/gdb.btrace/instruction_history.exp completed in 0 seconds

                === gdb Summary ===

Thanks,
-- 
Pedro Alves