| Message ID | 20210401102530.12030-8-alex.bennee@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | various fixes, pre-PR (check-tcg, gdbstub, gitlab) | expand |
On 4/1/21 3:25 AM, Alex Bennée wrote: > check_break("SHA1Init") > > - # check step and inspect values > + # Check step and inspect values. We do a double next after the > + # breakpoint as depending on the version of gdb we may step the > + # preamble and not the first actual line of source. > + gdb.execute("next") > gdb.execute("next") > val_ctx = gdb.parse_and_eval("context->state[0]") > exp_ctx = 0x67452301 This double next seems just as fragile. Why don't you just set the breakpoint where you want, which appears to be sha1.c:138? r~
On 4/2/21 9:41 AM, Richard Henderson wrote: > On 4/1/21 3:25 AM, Alex Bennée wrote: >> check_break("SHA1Init") >> - # check step and inspect values >> + # Check step and inspect values. We do a double next after the >> + # breakpoint as depending on the version of gdb we may step the >> + # preamble and not the first actual line of source. >> + gdb.execute("next") >> gdb.execute("next") >> val_ctx = gdb.parse_and_eval("context->state[0]") >> exp_ctx = 0x67452301 > > This double next seems just as fragile. Why don't you just set the breakpoint > where you want, which appears to be sha1.c:138? That said, make check-tcg has passed, completely, on my laptop for the first time in a year. So, Tested-by: Richard Henderson <richard.henderson@linaro.org> r~
Richard Henderson <richard.henderson@linaro.org> writes: > On 4/1/21 3:25 AM, Alex Bennée wrote: >> check_break("SHA1Init") >> - # check step and inspect values >> + # Check step and inspect values. We do a double next after the >> + # breakpoint as depending on the version of gdb we may step the >> + # preamble and not the first actual line of source. >> + gdb.execute("next") >> gdb.execute("next") >> val_ctx = gdb.parse_and_eval("context->state[0]") >> exp_ctx = 0x67452301 > > This double next seems just as fragile. Why don't you just set the > breakpoint where you want, which appears to be sha1.c:138? Ideally I'd want an unambiguous label that gdb would understand lest it got broken by someone tweaking the sha1.c code. I thought a function header would be better but it appears only marginally so. > > > r~
diff --git a/tests/tcg/multiarch/gdbstub/sha1.py b/tests/tcg/multiarch/gdbstub/sha1.py index 2bfde49633..423b720e6d 100644 --- a/tests/tcg/multiarch/gdbstub/sha1.py +++ b/tests/tcg/multiarch/gdbstub/sha1.py @@ -40,7 +40,10 @@ def run_test(): check_break("SHA1Init") - # check step and inspect values + # Check step and inspect values. We do a double next after the + # breakpoint as depending on the version of gdb we may step the + # preamble and not the first actual line of source. + gdb.execute("next") gdb.execute("next") val_ctx = gdb.parse_and_eval("context->state[0]") exp_ctx = 0x67452301
Depending on the version of gdb we may not execute the first line of SHA1Init when executing the first "next" command - instead just stepping over the preamble. As we don't actually care about the position of the PC after the steps and want to be sure the context->state[] has been loaded before we inspect it do a double next at the start. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> --- tests/tcg/multiarch/gdbstub/sha1.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)