Age | Commit message (Collapse) | Author |
|
The GDB testsuite has some tests which fail with the KFAIL status, which is
when there's a bugzilla report filed for that given failure.
Change-Id: I94d3cb2ed6aa1432a956c75ce1754a051ae7c7b2
|
|
Nearly all statuses have a "FOO → FOO" test, except for UNRESOLVED,
UNSUPPORTED and UNTESTED. For completeness, add them.
Change-Id: I16a5ce04003bba146b968581e0926034920824f8
|
|
The testsuite's sum files have a convention where a number is increased for
every test, but this isn't followed in tests-appear2.exp. It looks like a
copy/paste issue, so adjust.
Change-Id: I505e102f35432e9a7a346b1a50addbd0302cc4f1
|
|
I thought it was dirt on my screen, but it turns out it wasn't.
Change-Id: I4615d0cf7f984989718226eb75add3c4bd74b351
|
|
This will help report regressions or start bisections .
The .exp file name is still listed at the beginning of the section,
the reason it was only present there so far was to make reports more
compact.
This change showed that an .exp file didn't have the expected name in
sum-2.txt, so it is updated as needed, along with the expected
results.
Change-Id: I7571a59d291d77f6e7db51b598c1dc98e0c1ae7d
|
|
We count the number of occurrences of each test name (normally should
be either 0 or 1, but unfortunately there are some duplicates by
mistake sometimes). If this number is not consistent it means some
duplicates have appeared or disappeared.
Change-Id: I65e32e3703ea785511f87067aae67d53dec3bda7
|
|
Change-Id: I3787fcb314706648465211c60fff6a9d708c24e6
|
|
In the detailed report, we listed XFAIL->PASS and XFAIL-XPASS twice,
this patch removes this duplication.
Also change the addition order when computing $rtotal, to match the
output order (easier to read/maintain).
Change-Id: I40a98e0ef0502979a12034feebdbf1ca34f2330b
|
|
This has the same rationale as the "PASS disappears" case.
Change-Id: I8e6230dfa74064a1351c1b76f4e609b1c94e88b7
|
|
The new --hwdep-tests is similar to unstable-tests and provides a way
to specify a list of tests whose behaviour is considered as
hardware-dependent.
Such tests normally PASS is the hardware has the required feature,
they are skipped otherwise. This means that if such tests do not
always run the very same hardware, they appear/disappear from the
results.
To avoid noise in the results, this new flag allows to ignore such
cases.
It's OK for such tests to PASS, be UNSUPPORTED or absent from the
list. Other transitions are reported.
This patch implements TCWG-1421.
Change-Id: I25c125577ca795ea570342bcb95feef520096ff5
|
|
This makes it easier to reproduce regressions, avoiding the need to check
in the .sum which .exp was involved.
Change-Id: Iebd6f5ecf218037e372600f61168db8240ac7401
|
|
Change-Id: I3518a9d866770af2996c338309c5c4a9abc5e3a4
|