Re: [PATCH] gdb/testsuite: use with_timeout_factor in 2 more tests

[Jan Vrany <jan@vrany.io>]

On Mon, 2026-10-05 at 10:03 -0400, Simon Marchi wrote:
> On 10/5/26 5:52 AM, Jan Vrany wrote:
> > When running tests on some really slow targets, I got timeout in
> > gdb.base/funcargs.exp and gdb.base/completion.exp. The usual remedy I
> > use is to set timeout to higher value in site.exp.
> 
> I'd be curious to know which tests in those files exactly, see if we can
> improve either the test or GDB.

On my Vision5 RISC-V board I saw completion tests failing:

Running /home/jv/Projects/gdb/master/build/riscv64-linux-gnu/gdb/testsuite/../../../../gdb/testsuite/gdb.base/completion.exp ...
FAIL: gdb.base/completion.exp: complete 'p ' (timeout)
FAIL: gdb.base/completion.exp: complete 'info t foo' (timeout)
FAIL: gdb.base/completion.exp: complete 'info t' (timeout)
FAIL: gdb.base/completion.exp: complete 'info t ' (timeout)
FAIL: gdb.base/completion.exp: complete 'info asdfgh' (timeout)
FAIL: gdb.base/completion.exp: complete 'info asdfgh ' (timeout)
FAIL: gdb.base/completion.exp: complete 'info' (timeout)
FAIL: gdb.base/completion.exp: complete 'info ' (timeout)
FAIL: gdb.base/completion.exp: complete (2) 'info ' (timeout)
FAIL: gdb.base/completion.exp: complete 'help info wat' (timeout)
FAIL: gdb.base/completion.exp: complete 'p "break1' (timeout)
FAIL: gdb.base/completion.exp: complete 'p 'arg' (timeout)
FAIL: gdb.base/completion.exp: complete (2) 'p 'arg' (timeout)
FAIL: gdb.base/completion.exp: complete 'handle signal' (timeout)
FAIL: gdb.base/completion.exp: complete 'handle keyword' (timeout)
FAIL: gdb.base/completion.exp: complete help aliases (timeout)
FAIL: gdb.base/completion.exp: complete 'p no_var_named_this-arg' (timeout)
FAIL: gdb.base/completion.exp: complete (2) 'p no_var_named_this-arg' (timeout)
FAIL: gdb.base/completion.exp: complete (2) 'p no_var_named_this-' (timeout)
FAIL: gdb.base/completion.exp: complete 'p values[0].a' (timeout)
FAIL: gdb.base/completion.exp: complete 'p values[0] . a' (timeout)
FAIL: gdb.base/completion.exp: complete 'p &values[0] -> a' (timeout)
FAIL: gdb.base/completion.exp: completion of field in anonymous union

It think it is really just the CPU being slow by today's standards.

Interestingly, I cannot reproduce gdb.base/funcargs.exp failures on any
of my machines though I'm pretty sure I've seen them. Perhaps they were
overloaded or some NFS issue. Anyways, I think it would be better to
use with_timeout_factor there too just in case.

> 
> > However this does not help in case of aforementioned tests because they
> > set timeout to fixed value, regardless of whether the timeout has been
> > increased or not, causing them to fail on some of my machines.
> > 
> > This commit fixes the problem by using with_timeout_factor.
> > 
> > Since with_timeout_factor wraps large chunks of code, I decided not
> > to reformat (indent) it to make the patch easier to read and to
> > avoid "noise" when git-blaming.
> 
> I would suggest do the re-indent.  I find readability of current code
> more important than hypothetic git-blamability.  Plus, I believe that
> anyone doing any serious git-blaming should use a better tool than plain
> bit-blame, allowing to easily jump over uninteresting commits (for
> instance, I use "tig blame" and jumping over an uninteresting commit is
> one keystroke).

Fair point - will reformat.

Thanks!
> 
> > ---
> >  gdb/testsuite/gdb.base/completion.exp |  6 +++++-
> >  gdb/testsuite/gdb.base/funcargs.exp   | 11 +++++++----
> >  2 files changed, 12 insertions(+), 5 deletions(-)
> > 
> > diff --git a/gdb/testsuite/gdb.base/completion.exp b/gdb/testsuite/gdb.base/completion.exp
> > index 44cf649d5b6..f145f084999 100644
> > --- a/gdb/testsuite/gdb.base/completion.exp
> > +++ b/gdb/testsuite/gdb.base/completion.exp
> > @@ -61,7 +61,8 @@ if {![runto_main]} {
> >      return
> >  }
> > 
> > -set timeout 30
> > +with_timeout_factor 3 {
> > +
> >  gdb_test_no_output "set max-completions unlimited"
> > 
> >  gdb_test_no_output "complete print values\[0\].x." \
> > @@ -995,3 +996,6 @@ foreach_with_prefix spc { " " "" } {
> >  	    }
> >      }
> >  }
> > +
> > +# end of with_timeout_factor
> > +}
> > diff --git a/gdb/testsuite/gdb.base/funcargs.exp b/gdb/testsuite/gdb.base/funcargs.exp
> > index fdfdf07b78c..dd3240ba5e0 100644
> > --- a/gdb/testsuite/gdb.base/funcargs.exp
> > +++ b/gdb/testsuite/gdb.base/funcargs.exp
> > @@ -1031,13 +1031,15 @@ proc_with_prefix stepping_over_trampolines { } {
> >      }
> >  }
> > 
> > -set prev_timeout $timeout
> > +
> >  if {[istarget "mips*tx39-*"]} {
> > -    set timeout 300
> > +    set timeout_factor 30
> >  } else {
> > -    set timeout 60
> > +    set timeout_factor 6
> >  }
> 
> I would suggest getting rid of this special "mips*tx39-*" case.
> 
> LGTM with those fixed.
> 
> Approved-By: Simon Marchi <simon.marchi@efficios.com>
> 
> Simon