https://bugs.winehq.org/show_bug.cgi?id=43959
Bug ID: 43959 Summary: webservices/reader tests fail on arm Product: Wine Version: 2.19 Hardware: arm OS: Linux Status: NEW Keywords: download, testcase Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs@winehq.org Reporter: austinenglish@gmail.com Distribution: Debian
fixme:webservices:set_fpword not implemented reader.c:3733: Test failed: 0: got 80004001 fixme:webservices:set_fpword not implemented reader.c:3733: Test failed: 1: got 80004001 fixme:webservices:set_fpword not implemented reader.c:3733: Test failed: 2: got 80004001
Pretty obvious why: https://source.winehq.org/git/wine.git/blob/039d267b0925273197a9edcf7664c4a4...
3599 BOOL set_fpword( unsigned short new, unsigned short *old ) 3600 { 3601 #if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) 3602 unsigned short fpword; 3603 3604 __asm__ __volatile__( "fstcw %0" : "=m" (fpword) ); 3605 *old = fpword; 3606 fpword = new; 3607 __asm__ __volatile__( "fldcw %0" : : "m" (fpword) ); 3608 return TRUE; 3609 #else 3610 FIXME( "not implemented\n" ); 3611 return FALSE; 3612 #endif 3613 }
https://bugs.winehq.org/show_bug.cgi?id=43959
André H. nerv@dawncrow.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |nerv@dawncrow.de
--- Comment #1 from André H. nerv@dawncrow.de --- confirming: ... fixme:webservices:set_fpword not implemented reader.c:6263: Test failed: 51: got 80004001 reader.c:6388: Test marked todo: got 3 0008:reader: 3336 tests executed (46 marked as todo, 98 failures), 0 skipped.
https://bugs.winehq.org/show_bug.cgi?id=43959
André H. nerv@dawncrow.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |alexhenrie24@gmail.com
--- Comment #2 from André H. nerv@dawncrow.de --- https://source.winehq.org/patches/data/138483 https://source.winehq.org/patches/data/138484
reader and writer tests don't get down to zero, but that seems to be precision related
Adding Alex, as he is doing some precision related work for win64 tests atm it seems
reader.c:3734: Test failed: 55: got 7ff0000000000000 reader.c:3734: Test failed: 57: got 0
writer.c:388: Test failed: 2280: got 11 expected 10 writer.c:390: Test failed: 2280: got <t>:E-3</t> expected <t>1E-2</t> writer.c:388: Test failed: 2280: got 12 expected 11 writer.c:390: Test failed: 2280: got <t>-:E-3</t> expected <t>-1E-2</t> writer.c:390: Test failed: 2280: got <t>1.7976931348623156E+308</t> expected <t>1.7976931348623157E+308</t> writer.c:390: Test failed: 2280: got <t>-1.7976931348623156E+308</t> expected <t>-1.7976931348623157E+308</t>
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #3 from Hans Leidekker hans@meelstraat.net --- (In reply to André H. from comment #2)
https://source.winehq.org/patches/data/138483 https://source.winehq.org/patches/data/138484
reader and writer tests don't get down to zero, but that seems to be precision related
Adding Alex, as he is doing some precision related work for win64 tests atm it seems
reader.c:3734: Test failed: 55: got 7ff0000000000000 reader.c:3734: Test failed: 57: got 0
writer.c:388: Test failed: 2280: got 11 expected 10 writer.c:390: Test failed: 2280: got <t>:E-3</t> expected <t>1E-2</t> writer.c:388: Test failed: 2280: got 12 expected 11 writer.c:390: Test failed: 2280: got <t>-:E-3</t> expected <t>-1E-2</t> writer.c:390: Test failed: 2280: got <t>1.7976931348623156E+308</t> expected <t>1.7976931348623157E+308</t> writer.c:390: Test failed: 2280: got <t>-1.7976931348623156E+308</t> expected <t>-1.7976931348623157E+308</t>
Do the tests pass on ARM Windows?
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #4 from Alex Henrie alexhenrie24@gmail.com --- This does not look like a precision problem:
writer.c:390: Test failed: 2280: got <t>:E-3</t> expected <t>1E-2</t>
':' == '9' + 1, so it looks like an off-by-one error. We should fix this first.
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #5 from Hans Leidekker hans@meelstraat.net --- The code depends on the extra precision offered by the long double type on x86. It's an 80-bit quantity on x86 and 64-bit on ARM, AFAICT.
Before we make changes to the code I'd like to confirm that the current tests reflect ARM Windows behavior.
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #6 from Austin English austinenglish@gmail.com --- (In reply to Hans Leidekker from comment #5)
Before we make changes to the code I'd like to confirm that the current tests reflect ARM Windows behavior.
I just ordered a used Surface RT on ebay, so hopefully I can get more info on this in the coming days.
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #7 from Austin English austinenglish@gmail.com --- Created attachment 67185 --> https://bugs.winehq.org/attachment.cgi?id=67185 surface rt output
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #8 from Austin English austinenglish@gmail.com --- (In reply to Hans Leidekker from comment #5)
The code depends on the extra precision offered by the long double type on x86. It's an 80-bit quantity on x86 and 64-bit on ARM, AFAICT.
Before we make changes to the code I'd like to confirm that the current tests reflect ARM Windows behavior.
It passes on windows arm (wasn't really worth attaching, oops): 0050:reader: 3624 tests executed (0 marked as todo, 0 failures), 0 skipped.
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #9 from Hans Leidekker hans@meelstraat.net --- (In reply to Austin English from comment #8)
(In reply to Hans Leidekker from comment #5)
The code depends on the extra precision offered by the long double type on x86. It's an 80-bit quantity on x86 and 64-bit on ARM, AFAICT.
Before we make changes to the code I'd like to confirm that the current tests reflect ARM Windows behavior.
It passes on windows arm (wasn't really worth attaching, oops): 0050:reader: 3624 tests executed (0 marked as todo, 0 failures), 0 skipped.
There were changes in this code, does it still fail on Wine?
https://bugs.winehq.org/show_bug.cgi?id=43959
--- Comment #10 from Austin English austinenglish@gmail.com --- (In reply to Hans Leidekker from comment #9)
(In reply to Austin English from comment #8)
(In reply to Hans Leidekker from comment #5)
The code depends on the extra precision offered by the long double type on x86. It's an 80-bit quantity on x86 and 64-bit on ARM, AFAICT.
Before we make changes to the code I'd like to confirm that the current tests reflect ARM Windows behavior.
It passes on windows arm (wasn't really worth attaching, oops): 0050:reader: 3624 tests executed (0 marked as todo, 0 failures), 0 skipped.
There were changes in this code, does it still fail on Wine?
reader.c now passes in wine-5.8-230-g3bb824f988. Currently bisected the fix.
writer.c still fails: austin@PrawnOS:~/wine-git/dlls/webservices/tests$ WINEDEBUG=-all make writer.ok ../../../tools/runtest -q -P wine -T ../../.. -M webservices.dll -p webservices_test.exe.so writer && touch writer.ok writer.c:371: Test failed: 2272: got 11 expected 10 writer.c:373: Test failed: 2272: got <t>:E-3</t> expected <t>1E-2</t> writer.c:371: Test failed: 2272: got 12 expected 11 writer.c:373: Test failed: 2272: got <t>-:E-3</t> expected <t>-1E-2</t> writer.c:373: Test failed: 2272: got <t>1.7976931348623156E+308</t> expected <t>1.7976931348623157E+308</t> writer.c:373: Test failed: 2272: got <t>-1.7976931348623156E+308</t> expected <t>-1.7976931348623157E+308</t> writer.c:2316: Test failed: got 00000000 writer.c:2332: Test failed: got 00000000
Filed bug 49197 for that.
https://bugs.winehq.org/show_bug.cgi?id=43959
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Fixed by SHA1| |6d70b1a66e01315dedf27d65738 | |b8f050552d7ed Resolution|--- |FIXED
--- Comment #11 from Austin English austinenglish@gmail.com --- Fixed by 6d70b1a66e01315dedf27d65738b8f050552d7ed
https://bugs.winehq.org/show_bug.cgi?id=43959
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #12 from Alexandre Julliard julliard@winehq.org --- Closing bugs fixed in 5.9.
-
wine-bugs＠winehq.org -
WineHQ Bugzilla