March 2023 - Wine-Bugs - list.winehq.org

[Bug 54743] New: urlmon:url - The ftp test sometimes gets INET_E_OBJECT_NOT_FOUND on Windows

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54743 Bug ID: 54743 Summary: urlmon:url - The ftp test sometimes gets INET_E_OBJECT_NOT_FOUND on Windows Product: Wine Version: unspecified Hardware: x86-64 OS: Windows Status: NEW Severity: normal Priority: P2 Component: urlmon Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com urlmon:url - The ftp test sometimes gets INET_E_OBJECT_NOT_FOUND on Windows: url.c:1846: Test failed: binding failed: 800c0006, expected 00000000 url.c:1857: Test failed: res = 00002f4f, expected 00000000 url.c:3284: Test failed: expected OnProgress_SENDINGREQUEST url.c:3292: Test failed: expected OnProgress_MIMETYPEAVAILABLE url.c:3293: Test failed: expected OnProgress_BEGINDOWNLOADDATA url.c:3294: Test failed: expected OnProgress_ENDDOWNLOADDATA url.c:3306: Test failed: expected OnDataAvailable See https://test.winehq.org/data/patterns.html#urlmon:url Where 0x800c0006 == INET_E_OBJECT_NOT_FOUND This happens on Windows 8 to 11 about twice a month on average over all the VMs. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54742] The 64-bit advapi32:registry breaks the 32-bit test_redirection() in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54742 François Gouget <fgouget(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sven.wine(a)gmail.com -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54739] oleaut32:typelib - The 64-bit test_register_typelib_64() crashes in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54739 François Gouget <fgouget(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sven.wine(a)gmail.com Keywords| |regression Regression SHA1| |3f82f6ff59f4024f67e2d646883 | |cbea223fba84a --- Comment #1 from François Gouget <fgouget(a)codeweavers.com> --- A bisect shows that this started with the commit below: commit 3f82f6ff59f4024f67e2d646883cbea223fba84a Author: Sven Baars <sbaars(a)codeweavers.com> Date: Fri Mar 17 16:52:51 2023 +0100 kernelbase: Use open_key() to obtain any existing Wow6432node in create_key(). -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54742] The 64-bit advapi32:registry breaks the 32-bit test_redirection() in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54742 François Gouget <fgouget(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression Regression SHA1| |8b327eee5e7dc21c117515c6966 | |2d39693ada167 --- Comment #1 from François Gouget <fgouget(a)codeweavers.com> --- A bisect shows that these failures started with the commit that introduced them: commit 8b327eee5e7dc21c117515c69662d39693ada167 Author: Sven Baars <sbaars(a)codeweavers.com> Date: Wed Oct 19 14:52:42 2022 +0200 advapi32/tests: Copy Software\Classes tests from ntdll. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54738] msi:action - The 64-bit test_register_class_info() fails in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54738 François Gouget <fgouget(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sven.wine(a)gmail.com -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54738] msi:action - The 64-bit test_register_class_info() fails in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54738 François Gouget <fgouget(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression Regression SHA1| |3f82f6ff59f4024f67e2d646883 | |cbea223fba84a --- Comment #1 from François Gouget <fgouget(a)codeweavers.com> --- A bisect shows that this is caused by the commit below: commit 3f82f6ff59f4024f67e2d646883cbea223fba84a Author: Sven Baars <sbaars(a)codeweavers.com> Date: Fri Mar 17 16:52:51 2023 +0100 kernelbase: Use open_key() to obtain any existing Wow6432node in create_key(). -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54742] The 64-bit advapi32:registry breaks the 32-bit test_redirection() in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54742 François Gouget <fgouget(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Keywords| |source, testcase -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54742] New: The 64-bit advapi32:registry breaks the 32-bit test_redirection() in Wine

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54742 Bug ID: 54742 Summary: The 64-bit advapi32:registry breaks the 32-bit test_redirection() in Wine Product: Wine Version: unspecified Hardware: x86-64 OS: Linux Status: NEW Severity: normal Priority: P2 Component: advapi32 Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com Distribution: --- The 64-bit advapi32:registry breaks the 32-bit test_redirection() in Wine. To reproduce run the following commands: $ rm -rf ~/.wine # start from a fresh wineprefix $ ./wow64/wine64 wow64/dlls/advapi32/tests/x86_64-windows/advapi32_test.exe registry 2>&1 | grep -E '(Test failed|tests executed)' 0020:registry: 6309 tests executed (156 marked as todo, 0 as flaky, 0 failures), 2 skipped. $ ./wow32/wine wow32/dlls/advapi32/tests/i386-windows/advapi32_test.exe registry 2>&1 | grep -E '(Test failed|tests executed)' registry.c:2837: Test failed: RegOpenKeyExA failed: 2 registry.c:2842: Test failed: RegOpenKeyExA failed: 2 registry.c:2861: Test failed: 00000000: wrong value 32/64 registry.c:2863: Test failed: 00000200: wrong value 32/64 registry.c:2868: Test failed: 00000000: wrong value 64/0 registry.c:2869: Test failed: 00000100: wrong value 64/0 registry.c:2870: Test failed: 00000200: wrong value 64/0 registry.c:3168: Test failed: RegOpenKeyExA failed: 2 registry.c:3174: Test failed: RegOpenKeyExA failed: 2 registry.c:3180: Test failed: RegOpenKeyExA failed: 2 registry.c:3206: Test failed: found equals 0 registry.c:3208: Test failed: found equals 0 registry.c:3210: Test failed: found equals 0 registry.c:3212: Test failed: found equals 1 registry.c:3214: Test failed: wrong number of subkeys: 10 registry.c:3216: Test failed: found equals 1 registry.c:3218: Test failed: found equals 0 registry.c:3220: Test failed: wrong number of subkeys: 10 registry.c:3222: Test failed: found equals 0 0128:registry: 7658 tests executed (74 marked as todo, 0 as flaky, 19 failures), 2 skipped. See https://test.winehq.org/data/patterns.html#advapi32:registry The reason why this failure is only visible with the fg-deb64-wow32 is because this is the only test configuration that runs the 32-bit tests in the same wineprefix as the 64-bit tests. What this shows is that advapi32:registry does not correctly cleans up after itself which has the potential for breaking any test that follows (not just 32-bit ones). -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 54741] New: integer overflow in get_dib_stride / NtGdiCreateBitmap

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=54741 Bug ID: 54741 Summary: integer overflow in get_dib_stride / NtGdiCreateBitmap Product: Wine Version: 8.4 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: gdi32 Assignee: wine-bugs(a)winehq.org Reporter: felix-wine(a)fefe.de Distribution: --- Here's the source code of NtGdiCreateBitmap: 97 HBITMAP WINAPI NtGdiCreateBitmap( INT width, INT height, UINT planes, 98 UINT bpp, const void *bits ) 99 { Having INT for width and height is already bad, but can't be helped because you are implementing a crappy API. 105 if (width > 0x7ffffff || height > 0x7ffffff) 106 { 107 RtlSetLastWin32Error( ERROR_INVALID_PARAMETER ); 108 return 0; 109 } 110 111 if (!width || !height) 112 return 0; 113 114 if (height < 0) 115 height = -height; 116 if (width < 0) 117 width = -width; After this the value of width and height is 1..0x7ffffff (7 digits, not 8). bpp is validated to be at most 32. 140 dib_stride = get_dib_stride( width, bpp ); 141 size = dib_stride * height; 142 /* Check for overflow (dib_stride itself must be ok because of the constraint on bm.bmWidth above). */ 143 if (dib_stride != size / height) Here's the code of get_dib_stride: 282 static inline int get_dib_stride( int width, int bpp ) 283 { 284 return ((width * bpp + 31) >> 3) & ~3; 285 } width can be 0x7ffffff, bpp can be 32. width*bpp is then 0xffffffe0 which looks at first glance like no overflow happened, but the type is still int, i.e. signed 32-bit, i.e. max value is 0x7fffffff (8 digits), max bpp to trigger that would be 16 not 32. We are right-shifting afterwards, so we ought to be fine? Nope, because we are operating on signed ints, so it's an arithmetic right shift, not a logical one. The & does not remove the sign bit. So what happens here is that get_dib_stride can return a negative value. OK but NtGdiCreateBitmap checks for overflow and will notice something is wrong. Not really. If width=height=0x7ffffff and bpp=32 then get_dib_stride=-4. Then size is -4*0x7ffffff promoted to size_t, or 0xe0000004. The overflow check does this: 143 if (dib_stride != size / height) size/0x7ffffff is 0xfffffffc. dib_stride is -4, promoted to size_t for the comparison is also 0xfffffffc. Also: On the way we had undefined behavior. The compiler would be allowed to do all kinds of nasty things. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023

[Bug 53269] uiautomationcore:uiautomation fails on Windows 10 1909

by WineHQ Bugzilla

https://bugs.winehq.org/show_bug.cgi?id=53269 Connor McAdams <conmanx360(a)gmail.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |conmanx360(a)gmail.com --- Comment #1 from Connor McAdams <conmanx360(a)gmail.com> --- This should now be fixed by commit 5c41987709a861d088c887a723aaf1779b1b9272 , the latest testbot run seems to confirm it is. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

March 28, 2023