The sim_init function was called by runtest for each test when --tool
was set to sim. When we changed to --tool '' to collapse the testsuite
dir, the init function was no longer called on every test. However, it
was still being called explicitly by config/default.exp. It's not clear
why that explicit call ever existed since, in the past, it meant it was
redundant.
Lets drop the single sim_init call in config/default.exp and move it out
to all our tests. This replicates the runtest behavior so we can setup
variables on a per-test basis which allows us to recollapse the sim_path
logic back. We'll also leverage this in the future for toolchain setup.
Also add a few comments clarifying the overall runtime behavior.
If the cris sim hasn't been built yet, trying to run its testsuite
will hang indefinitely. The common sim APIs already have this, so
copy it over to the cris forks of the test+run functions.
The tests assume that the cwd is the objdir directory and write its
intermediates to there all the time. When using runtest's --objdir
setting though, this puts the files in the wrong place. This isn't
a big problem currently as we never change --objdir, but in order to
support parallel test execution, we're going to start setting that
option, so clean up the code ahead of time.
We also have to tweak some of the cris tests which were making
assumptions about the argv[0] value.
Now that all the other toolchain settings have been renamed to match
the dejagnu settings of XXX_FOR_TARGET, rename global_sim_options to
SIMFLAGS_FOR_TARGET too.
Only a few tests actually use global_ld_options, but we can replace the
sim-specific settings with the dejagnu common LDFLAGS_FOR_TARGET and get
the same result.
Only a few tests actually use global_as_options, but we can replace the
sim-specific settings with the dejagnu common ASFLAGS_FOR_TARGET and get
the same result.
Nothing in the testsuite is using this setting, so let's drop it.
Any code that wants to set compiler flags can use CFLAGS_FOR_TARGET
instead to get the same effect.
These haven't been used in over 20 years. The sim testsuite used to
run these tools itself directly, but back in ~1999 it switched to the
dejagnu helpers (e.g. target_assemble & target_link), and the dejagnu
logic only utilizes XXX_FOR_TARGET variables. Punt them here to avoid
confusion with dead code.
The v850 testsuite code has been testing the $opt variable, but this
was never actually set anywhere globally or v850-specific. Instead,
this was a random variable leaking out of the sh testsuite code. As
far as I can tell, it has always been this way. That means the code
only ever tested the v850 cpu target (which is the default).
This failure can be easily seen in practice by running the v850 code
in isolation and seeing it crash:
$ runtest v850/allinsns.exp
...
Running target unix
Using /usr/share/dejagnu/baseboards/unix.exp as board description file for target.
Using /usr/share/dejagnu/config/unix.exp as generic interface file for target.
Using ../../../sim/testsuite/config/default.exp as tool-and-target-specific interface file.
WARNING: Assuming target board is the local machine (which is probably wrong).
You may need to set your DEJAGNU environment variable.
Running ../../../sim/testsuite/v850/allinsns.exp ...
ERROR: tcl error sourcing ../../../sim/testsuite/v850/allinsns.exp.
ERROR: tcl error code TCL LOOKUP VARNAME opt
ERROR: can't read "opt": no such variable
while executing
"switch -regexp -- $opt {
Backing up a bit, the reason for this logic in the first place is
because the common sim testsuite code makes an assumption about the
assembler options with cpu_option -- the option and its value are
always separated by an =. This is not the case with v850. So tweak
the core sim logic a bit to support omitting the = so that we can
switch v850 to the standard all_machs setting and avoid opt entirely.
The upstream GCC tester has showed spurious execution failures on the
H8 target for the H8/SX multilibs. I suspected memory corruption or an
uninitialized variable early as the same binary would sometimes work and
sometimes it got the wrong result. Worse yet, the point where the test
determined it was getting the wrong result would change.
Because it only happened on the H8/SX variant I was able to zero in on
the "mova" support and the "short form" of those instructions in particular.
As the code stands it checks if code->op3.type == 0 to try and identify cases
where op3 wasn't filled in and thus we've got the short form of the mova
instruction.
But for the short-form of those instructions we never set any of the "op3"
data structure. We get whatever was lying around -- it's usually zero and
thus things usually work, but if the stale data was nonzero, then we'd
fail to recognize the instruction as a short-form and fail to set up the
various fields appropriately.
I initially initialized the op3.type field to zero, but didn't like that
because it was inconsistent with how other operands were initialized.
Bringing consistency meant using -1 as the initializer value and adjusting
the check for short form mova appropriately.
I've had this in the upstream GCC tester for perhaps a year at this point
and haven't seen any of the intermittent failures again.
Instead of always using target_alias as a prefix on the name, use
program_transform_name instead so that the library is scoped in the
same way as the run program.
The common framework supports $srcdir & $subdir replacements already,
so replace the custom @srcdir@ logic with those. Since the replace
happens in slurp_options that cris already uses, we don't have any
logic to port over there. We have to duplicate that into the cris
slurp_rv helper though.
This tag is used to force tests to be built dynamically (i.e. without
-static linking). This is because cris-sim.exp in dejagnu turns on
static linking in ldflags.
The default configs and runtest flags shouldn't load these boards.
If these settings are still needed, we should figure out a different
way of suppressing the stock settings wholesale. We want these to
all pass out of the box with little to no configuration so that they
can run in a multitarget build.
With dropping "dynamic", it'll be easier to merge the custom cris
test logic with the common sim test logic.
On gcc-12 build fails as:
sim/cr16/interp.c: In function 'lookup_hash':
sim/cr16/interp.c:89:25: error:
the comparison will always evaluate as 'true'
for the address of 'mnimonic' will never be NULL [-Werror=address]
89 | if ((h->ops->mnimonic != NULL) &&
| ^~
'mnimonic' is a sharr array within ops. It can never be NULL.
While at it renamed 'mnimonic' to 'mnemonic'.
Support the same syntax as other common utilities where env vars can
be specified before the program to be run without an explicit option.
This behavior can be suppressed by using the -- marker.
Provide explicit control over the program's environment with the
basic set/unset/clear options. These are a bit clunky to use,
but they're functional.
The env set operation is split out into a separate function as it'll
be used in the next commit.
With these in place, we can adjust the custom cris testsuite to use
the now standard options and not its one-off hack.
Now that we have access to the argv & envp strings, finish implementing
these syscalls. Delete unused variables, fix tbuf by incrementing the
pointer instead of setting to the length, and make sure we don't write
more data than the bufsize says is available.
We've been passing the environment strings to sim_create_inferior,
but most ports don't do anything with them. A few will use ad-hoc
logic to stuff the stack for user-mode programs, but that's it.
Let's formalize this across the board by storing the strings in the
normal sim state. This will allow (in future commits) supporting
more functionality in the run interface, and to unify some of the
libgloss syscalls.
The point of this test is to just make sure the usage text is shown,
not the exact details of the usage text. So shorten the output test
to match the beginning. This fixes breakage when the output changed
slightly to include [--].
The new argv processing code assumed that we were always passed a
command line. If we weren't, make sure we don't crash before we
get a chance to output an error message about incorrect usage.
The #dest field used in the cris testsuite is a bit of hack to set the
argv[0] for the tests to read out later on. Now that the sim has an
option to set argv[0] explicitly, we don't need this custom field, so
let's drop it to harmonize the testsuites a little.
We default argv[0] to the program we run which is a standard *NIX
convention, but sometimes we want to be able to control the argv[0]
setting independently (especially for programs that inspect argv[0]
to change their behavior or output). Add an option to control it.
We use the program argv to both find the program to run (argv[0]) and
to hold the arguments to the program. Most of the time this is fine,
but if we want to let programs specify argv[0] independently (which is
possible in standard *NIX programs), this double duty doesn't work.
So let's split the path to the program to run out into a separate
field by itself. This simplifies the various sim_open funcs too.
By itself, this code is more of a logical cleanup than something that
is super useful. But it will open up customization of argv[0] in a
follow up commit. Split the changes to make it easier to review.
Set the mach to the right value all the time, and update xfail to
say the test fails on all targets. WIth multitarget testing, the
idea of target here doesn't make much sense.
In <=gcc-7 versions, -fstrict-overflow is enabled by default, and that
triggers warnings in this code that relies on integer overflows to test
for carries. Change the logic to test against the limit directly.
Since gcc dropped support for a.out starting with 4.4.0 in 2009, it's
been impossible to verify this code actually still works. Since it
crashes in ld, and it uses a config option that no other tests uses
and we want to remove, drop the test to avoid all the trouble.
Outputting an extra space broke a cris test. Change the workaround
to use %s with an empty string to avoid the compiler warning but not
output an extra space.
There's no need to sync testutils.inc with remote hosts. The one
we have in the source tree is all we need and only thing we test.
Delete it to simplify.
Objcopy was used to create a binary file of just the executable code
since the environment requires code to based at address 0. We can
accomplish the same thing with the -Ttext=0 flag, so switch to that
to get rid of custom logic.
When I refactored this by scoping it to sim-frv-xxx in commit
e7954ef5e5 ("sim: frv: scope the
unique configure flag"), I changed the default from off to on.
While the feature is nice for developers, it breaks a bunch of
tests which aren't expecting this extra output. So flip it back
to off by default.
Switch from the centralized list in the exp file to each test declaring
its own requirements which they're already (mostly) doing. This will
increase coverage slightly by running more tests in more configurations
since the hardcoded exp list was a little out of date.
We have to mark the psh* tests as shdsp only (to match what the exp
file was doing), mark the fsca & fsrra tests as failing (since they
weren't even being run by the exp file), and to fix the expected
output & status of the fail test.
Move some unused funcs under existing #if 0 protection, mark a few
local funcs as static, and add missing prototypes for the rest which
are used from other files. This fixes all the fatal warnings in the
mloop files so we can turn -Werror on here fully.
On LLP64 targets where sizeof(long) != sizeof(void*), this code fails:
sim/sh/interp.c:704:24: error: cast from pointer to integer of different size -Werror=pointer-to-int-cast]
704 | do { memstalls += ((((long) PC & 3) != 0) ? (n) : ((n) - 1)); } while (0)
| ^
Since this code simply needs to check alignment, cast it using uintptr_t
which is the right type for this.
Casting 0 to a pointer via (long *) doesn't work on LLP64 targets:
error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
It's also unnecessary here. We can simply pass NULL like every other
bit of code does.
The code assumes that all _WIN32 targets are the same and can
define isnan to _isnan. For mingw targets, they provide an isnan
define already, so no need for the fallback here.
Some targets (like cygwin) will export page size defines that clash
with our local usage here. Undefine the system one to fix building
for these targets.
