https://bugs.winehq.org/show_bug.cgi?id=52447
Bug ID: 52447 Summary: Regression: .NET framework 2 installer hangs Product: Wine Version: 7.0 Hardware: x86-64 OS: FreeBSD Status: NEW Severity: normal Priority: P2 Component: ntdll Assignee: wine-bugs@winehq.org Reporter: damjan.jov@gmail.com
Shortly after Wine 6.11:
winetricks dotnet20
hangs while installing System.Windows.Forms. A git bisect gives:
fed7e7a78b4bea183d0cbb107953e275f17baa4f is the first bad commit commit fed7e7a78b4bea183d0cbb107953e275f17baa4f Author: Alexandre Julliard julliard@winehq.org Date: Mon Jun 21 15:57:26 2021 +0200
ntdll: Simulate a syscall return when starting a thread.
Signed-off-by: Alexandre Julliard julliard@winehq.org
dlls/ntdll/unix/server.c | 2 +- dlls/ntdll/unix/signal_arm.c | 65 ++++++++++--------------- dlls/ntdll/unix/signal_arm64.c | 70 +++++++++++---------------- dlls/ntdll/unix/signal_i386.c | 102 ++++++++++++++++++---------------------- dlls/ntdll/unix/signal_x86_64.c | 91 +++++++++++++++-------------------- dlls/ntdll/unix/thread.c | 2 +- dlls/ntdll/unix/unix_private.h | 2 +- 7 files changed, 138 insertions(+), 196 deletions(-)
https://bugs.winehq.org/show_bug.cgi?id=52447
Damjan Jovanovic damjan.jov@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |dotnet, regression Regression SHA1| |fed7e7a78b4bea183d0cbb10795 | |3e275f17baa4f CC| |julliard@winehq.org
--- Comment #1 from Damjan Jovanovic damjan.jov@gmail.com --- CC author, add keywords and regression SHA1.
https://bugs.winehq.org/show_bug.cgi?id=52447
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Regression: .NET framework |.NET framework 2 installer |2 installer hangs |hangs
https://bugs.winehq.org/show_bug.cgi?id=52447
Anastasius Focht focht@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|.NET framework 2 installer |64-bit .NET framework 2.0 |hangs |installer hangs while | |generating/installing | |native images of | |'System.Windows.Forms' | |assembly into GAC Keywords| |download CC| |focht@gmx.net URL| |https://web.archive.org/web | |/20060509045320/https://dow | |nload.microsoft.com/downloa | |d/a/3/f/a3f1bf98-18f3-4036- | |9b68-8e6de530ce0a/NetFx64.e | |xe
--- Comment #2 from Anastasius Focht focht@gmx.net --- Hello folks,
I can't reproduce this. The 64-bit .NET Framework 2.0 installer works fine w.r.t. assembly/native image installation into GAC. It definitely takes some time for native image generation for each assembly but no indefinite hang of any ngen worker or ngen service. They terminate successfully.
Might be FreeBSD thing.
Adding stable download link via Internet Archive.
https://web.archive.org/web/20060509045320/https://download.microsoft.com/do...
https://www.virustotal.com/gui/file/7ea86dca8eeaedcaa4a17370547ca2cea9e9b677...
$ sha1sum NetFx64.exe e59cca309463a5d98daeaada83d1b05fed5126c5 NetFx64.exe
$ du -sh NetFx64.exe 46M NetFx64.exe
$ wine --version wine-7.0-119-gc09a5da1575
Regards
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #3 from Damjan Jovanovic damjan.jov@gmail.com --- (In reply to Anastasius Focht from comment #2)
Hello folks,
I can't reproduce this. The 64-bit .NET Framework 2.0 installer works fine w.r.t. assembly/native image installation into GAC. It definitely takes some time for native image generation for each assembly but no indefinite hang of any ngen worker or ngen service. They terminate successfully.
Thank you.
Might be FreeBSD thing.
Must be. I can't reproduce it on Ubuntu either.
https://bugs.winehq.org/show_bug.cgi?id=52447
Alex S iwtcex@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |iwtcex@gmail.com
--- Comment #4 from Alex S iwtcex@gmail.com --- Maybe it's finally time to fix signal numbers (see i386_trap_code) on amd64.
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #5 from Damjan Jovanovic damjan.jov@gmail.com --- (In reply to Alex S from comment #4)
Maybe it's finally time to fix signal numbers (see i386_trap_code) on amd64.
Unfortunately that doesn't help, and I am only using the 32 bit Wine for this.
I've sent through a patch that fixes those trap codes anyway, for the sake of any other apps: https://source.winehq.org/patches/data/224351
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #6 from Alex S iwtcex@gmail.com ---
Unfortunately that doesn't help, and I am only using the 32 bit Wine for this.
Curious. Is that a PE (mingw) or non-PE build? Do you observe the same behavior with WINEDEBUG=+seh?
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #7 from Damjan Jovanovic damjan.jov@gmail.com --- (In reply to Alex S from comment #6)
Unfortunately that doesn't help, and I am only using the 32 bit Wine for this.
Curious. Is that a PE (mingw) or non-PE build? Do you observe the same behavior with WINEDEBUG=+seh?
On Linux both ELF and PE builds work. On FreeBSD we only have the ELF build, which breaks. WINEDEBUG='+seh' gives lots of output, ending with this printed endlessly while the CPU core is 100% busy:
07b4:trace:seh:handle_syscall_fault code=c0000005 flags=0 addr=0x41afebc ip=041afebc tid=07b4 07b4:trace:seh:handle_syscall_fault info[0]=00000001 07b4:trace:seh:handle_syscall_fault info[1]=c0000005 07b4:trace:seh:handle_syscall_fault eax=c0000005 ebx=041afee0 ecx=7bc3341c edx=00000000 esi=00000000 edi=00000000 07b4:trace:seh:handle_syscall_fault ebp=041afec8 esp=041cfbb4 cs=0033 ds=003b es=003b fs=0013 gs=001b flags=00010246 07b4:trace:seh:handle_syscall_fault returning to user mode ip=7bc3341c ret=c0000005
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #8 from Alex S iwtcex@gmail.com --- (In reply to Damjan Jovanovic from comment #7)
I can't consistently reproduce the hang with 32-bit Wine. A couple times it looked like that, but it's mostly working. I'm using Ryzen 5600X, if that makes any difference.
On FreeBSD we only have the ELF build
I'm pretty sure, per FreeBSD bureaucracy rules, the patch I've uploaded to the FreeBSD bug tracker awaits your approval.
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #9 from Damjan Jovanovic damjan.jov@gmail.com --- (In reply to Alex S from comment #8)
(In reply to Damjan Jovanovic from comment #7)
I can't consistently reproduce the hang with 32-bit Wine. A couple times it looked like that, but it's mostly working. I'm using Ryzen 5600X, if that makes any difference.
Still 100% reproducible for me on Wine 7.3 with the PE build too. A 1 hour wait isn't a case of "looked like that". LIBGL_ALWAYS_INDIRECT=1 and using the wrong LIBGL_DRIVERS_PATH had no effect, and 3D graphics don't appear to be used.
https://bugs.winehq.org/show_bug.cgi?id=52447
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|--- |NOTOURBUG Status|NEW |RESOLVED
--- Comment #10 from Alexandre Julliard julliard@winehq.org --- Assuming FreeBSD bug.
https://bugs.winehq.org/show_bug.cgi?id=52447
Damjan Jovanovic damjan.jov@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|NOTOURBUG |FIXED
--- Comment #11 from Damjan Jovanovic damjan.jov@gmail.com --- This works now: "winetricks dotnet20" succeeds every time, and finishes much faster.
Wine 10.0-rc3, compiled by GCC 14 and cross-compiled to PE by Clang 18, with no --enable-archs, on FreeBSD 14.1.
Not sure what fixed it, but resolving fixed.
https://bugs.winehq.org/show_bug.cgi?id=52447
Damjan Jovanovic damjan.jov@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Fixed by SHA1| |871f333d55c2c472014a861bfdd | |256c1f497dd40
--- Comment #12 from Damjan Jovanovic damjan.jov@gmail.com --- (In reply to Alexandre Julliard from comment #10)
Assuming FreeBSD bug.
It wasn't a FreeBSD bug, because this is what fixed it. FreeBSD simply revealed the Wine bug 15 months before it was apparent on Linux:
871f333d55c2c472014a861bfdd256c1f497dd40 is the first working commit commit 871f333d55c2c472014a861bfdd256c1f497dd40 Author: Alexandre Julliard julliard@winehq.org Date: Tue Apr 25 15:52:26 2023 +0200
ntdll: Use a separate memory allocation for the kernel stack.
dlls/ntdll/unix/thread.c | 35 ++++++++++++++++------------------- dlls/ntdll/unix/unix_private.h | 2 +- dlls/ntdll/unix/virtual.c | 38 +++++++++++--------------------------- 3 files changed, 28 insertions(+), 47 deletions(-)
https://bugs.winehq.org/show_bug.cgi?id=52447
--- Comment #13 from Damjan Jovanovic damjan.jov@gmail.com --- And for even older versions of Wine, this patch also needs to be backported, for .NET 2 to install on FreeBSD:
commit df5f9b7d2b5395c1a537981b802b804b9f93b8ae Author: Jinoh Kang jinoh.kang.kr@gmail.com Date: Wed Mar 22 01:36:45 2023 +0900
ntdll: Don't report user (PE) stack via pthread_attr_setstack().
https://bugs.winehq.org/show_bug.cgi?id=52447
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #14 from Alexandre Julliard julliard@winehq.org --- Closing bugs fixed in 10.0-rc4.
-
WineHQ Bugzilla