<div dir="ltr">That's weird.<div><br></div><div>All of the LLVM source files I have from git.exe have "Read & Execute" permission. If those aren't showing execute permission for you under Cygwin then we need to understand how Cygwin is mapping Windows file permissions to the Unix model before we can determine what we would need to change in LLVM to get the same outcome.</div><div><br></div><div>Can you use the Window's File Explorer properties page to compare the security settings of two source files, one that shows 644 in Cywin and one that shows 755 after being re-written by clang-format?</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Apr 20, 2020 at 9:32 AM Chris Tetreault <<a href="mailto:ctetreau@quicinc.com">ctetreau@quicinc.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div lang="EN-US">
<div class="gmail-m_-8987017097137136193WordSection1">
<p class="MsoNormal">I’m using Cygwin to interact with the source tree. My sources (I’m working on LLVM itself) were created by other developers (and by extension, created by git.exe on my machine), but a quick expirement of trying “touch foo” shows that Cygwin
creates files with 644 mode. Finding the file I created in explorer and checking the properties shows it has the following NTFS permissions:<u></u><u></u></p>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal">My user: Read, Write<u></u><u></u></p>
<p class="MsoNormal">Domain users: Read<u></u><u></u></p>
<p class="MsoNormal">Everyone: Read<u></u><u></u></p>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal">I suppose “whatever Cygwin does” might serve as a good model of how me might meaningfully translate unix-style permissions to windows. I’ve created a bug for this (45619), I can update it with this suggestion.<u></u><u></u></p>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal">As for the harmlessness of a source file being committed with executable permissions, I think it’s a security risk. An attacker can know that a program creates files with executable permissions. They can presumably trick llvm into producing
a script that they can run on a target machine. Even if you disagree that this is a big deal, I’ve worked on projects previously where you would get dinged in code review for setting executable permissions on source files. It would be hugely annoying to have
to remember to fixup the permissions every time you invoke clang-format.<u></u><u></u></p>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal">Thanks,<u></u><u></u></p>
<p class="MsoNormal"> Christopher Tetreault<u></u><u></u></p>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal"><b>From:</b> Adrian McCarthy <<a href="mailto:amccarth@google.com" target="_blank">amccarth@google.com</a>> <br>
<b>Sent:</b> Monday, April 20, 2020 8:36 AM<br>
<b>To:</b> Chris Tetreault <<a href="mailto:ctetreau@quicinc.com" target="_blank">ctetreau@quicinc.com</a>><br>
<b>Cc:</b> LLVM Dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>>; <a href="mailto:cfe-dev@lists.llvm.org" target="_blank">cfe-dev@lists.llvm.org</a><br>
<b>Subject:</b> [EXT] Re: [llvm-dev] clang-format sets executable permission on windows (openNativeFile ignores mode on Windows)<u></u><u></u></p>
<p class="MsoNormal"><u></u> <u></u></p>
<div>
<p class="MsoNormal">I don't claim to understand NTFS permissions fully, but this mostly sounds like a problem of how the NTFS permissions are presented as a Unix-style mode.<u></u><u></u></p>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">When you create a new file without specifying explicit permissions (as most tools do), Windows (or NTFS) grants the new file the same permissions as the folder that contains it. Folders generally have the "Read & Execute" permission, since
that's what lets a user navigate the filesystem. Whatever tool you're using to translate Windows/NTFS permissions into a Unix-style mode is probably showing x when the file has R&E.<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">I'm curious how your source files were originally created without R&E.<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">On something like a text file or a source file, R&E is common and harmless.<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">Common: If I create a text file from Explorer or Notepad or Visual Studio, that file gets R&E. I have source files that have never been touched by clang-format, and they have the same permissions, including R&E, as the ones that have.<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">Harmless: Since text files aren't executable, having R&E doesn't grant anything beyond Read. A possible exception might be batch files (.BAT). Does the command interpreter check R&E for those? I don't know offhand. If it does, would
you want to force the user to change permissions of a .BAT file they had just written in a text editor before they try it?<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">That said, I don't entirely understand the permission model has both "Read" and "Read & Execute". I'd guess that it's because having orthogonal "Read" and "Execute" permissions would allow a nonsense state that marks a file as executable
but not readable.<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">The flags like FILE_GENERIC_READ and FILE_GENERIC_EXECUTE don't map one-to-one with the Windows/NTFS file permissions. Rather, they are mostly used to specify the type of access a particular operation needs (to a call like CreateFileW,
which doesn't necessarily create a file but often opens one instead). The system checks the requested access against what's allowed by the permissions granted in the file's security descriptor (as well as types of shared access allowed by others who already
have the file open).<u></u><u></u></p>
</div>
</div>
<p class="MsoNormal"><u></u> <u></u></p>
<div>
<div>
<p class="MsoNormal">On Fri, Apr 17, 2020 at 4:53 PM Chris Tetreault via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>> wrote:<u></u><u></u></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0in 0in 0in 6pt;margin:5pt 0in 5pt 4.8pt">
<div>
<div>
<p class="MsoNormal">Hi,<u></u><u></u></p>
<p class="MsoNormal"> <u></u><u></u></p>
<p class="MsoNormal"> I’m having an issue where clang-format is setting the executable bit on all source files it modifies when using the -i parameter. I spent some time troubleshooting
this issue today, and I found that clang-format create a new temporary file, writes the formatted source into that file, then copies it over the old file. Deep in the bowels of openNativeFile in lib/Support/Windows/Path.inc, in openNativeFileInternal, CreateFileW
is called with a SECURITY_ATTRIBUTES with lpSecurityDescriptor = nullptr. The result of this is you get a new file with the default permissions based on whatever NTFS decides to do. On my machine, this ends up being a file with 755 mode. This is happening
because the mode parameter to openNativeFile is unused. This issue occurs in clang-format 9, and on HEAD of master.<u></u><u></u></p>
<p class="MsoNormal"> <u></u><u></u></p>
<p class="MsoNormal"> I spent some time thinking about how to improve this state of affairs. I feel like just letting files modified by clang-format get their permissions changed severely
limits the convenience of the tool. Just some thoughts I had:<u></u><u></u></p>
<p class="MsoNormal"> <u></u><u></u></p>
<p class="gmail-m_-8987017097137136193gmail-m8137147084588757416msolistparagraph" style="margin-left:0.75in">
<u></u><span>1.<span style="font:7pt "Times New Roman"">
</span></span><u></u>Would it be so terrible if files created by openNativeFile on windows queried the default SECURITY_ATTRIBUTES and stripped executable permissions? Does anything actually need llvm to produce files with executable permissions? I did a
quick search but can’t find anywhere in the codebase that actually sets a value for mode.<u></u><u></u></p>
<p class="gmail-m_-8987017097137136193gmail-m8137147084588757416msolistparagraph" style="margin-left:0.75in">
<u></u><span>2.<span style="font:7pt "Times New Roman"">
</span></span><u></u>How would we even map unix-style permissions to windows? I see in the MS docs that there are coarse-grained permission types that map to unix permissions (FILE_GENERIC_READ, FILE_GENERIC_WRITE, and FILE_GENERIC_EXECUTE). For file creation,
the current user could be the owner. Perhaps all groups the user is a member of could get the group permissions, and maybe Authenticated Users for other?<u></u><u></u></p>
<p class="gmail-m_-8987017097137136193gmail-m8137147084588757416msolistparagraph" style="margin-left:0.75in">
<u></u><span>3.<span style="font:7pt "Times New Roman"">
</span></span><u></u>On my previous project, we used clang-format, and I never had this issue. I was using a very old version though, so I don’t know if my configuration is just different, or if this behavior changed at some point<u></u><u></u></p>
<p class="MsoNormal"> <u></u><u></u></p>
<p class="MsoNormal">Thanks,<u></u><u></u></p>
<p class="MsoNormal"> Christopher Tetreault<u></u><u></u></p>
</div>
</div>
<p class="MsoNormal">_______________________________________________<br>
LLVM Developers mailing list<br>
<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a><br>
<a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><u></u><u></u></p>
</blockquote>
</div>
</div>
</div>
</blockquote></div>