[PATCH 0/7] MIPS+testsuite: BLEZL stepping fix and associated test infrastructure

"Maciej W. Rozycki" <[email protected]> Thu, 23 Jul 2026 19:52:42 +0100 (BST)
Newsgroups gmane.comp.gdb.patches
Message-ID <[email protected]>
Hi,

 This patch series is based on a buried bug fix extracted from a change 
submitted to add MIPSr6 support, most recently posted at 
<https://inbox.sourceware.org/gdb-patches/[email protected]/>, 
that corrects the handling of single-stepping through the BLEZL MIPS II 
instruction.

 This fix begged for test coverage, as indeed does all the MIPS software 
single-stepping infrastructure, so I have created a bunch of test cases 
roughly based on the skeleton already provided with said MIPSr6 change, 
however with extra feature checks to correctly enable individual tests 
based on the target configuration chosen for the testsuite invocation and 
target board's compatibility with the ISA required for test execution.

 This in turn revealed a repeated pattern of feature check code across 
numerous procedures in testsuite/lib/gdb.exp.  Rather than adding a bunch 
more, I chose to factor out the repeated pattern and build the new feature 
checks around it.  For illustration purposes I have converted the existing 
`powerpc*-*-*' target feature checks I could verify with my POWER9 test 
system.

 Consequently this set of 7 changes has resulted, comprised of the new 
generic procedure in 1/7, followed by the `powerpc*-*-*' updates in 2/7, 
and then the MIPS changes in 3/7 through 7/7.  Only MIPS I, MIPS II and 
MIPS16 CPU branches are covered at this point.  See individual change 
descriptions for details.  Authorship attributions are on the best 
knowledge basis.

 This patch set has been regression-tested with a `powerpc64le-linux-gnu' 
native configuration using POWER9 hardware and with the `mips-linux-gnu' 
and `mipsel-linux-gnu' targets, using real MIPS I, MIPS III and MIPS32r2 
hardware, the latter with regular MIPS and MIPS16 test configurations.  
Detailed logs were examined for correctness with respect to the tests 
affected as well.

 I request approval for the non-MIPS parts of the patch series; I can 
self-approve the MIPS bits, though obviously I'm happy to address any 
feedback.

 Company copyright assignment with FSF is currently being sorted.

  Maciej