Re: path length limit in seatbelt/sandbox?
site_archiver@lists.apple.com Delivered-To: darwin-kernel@lists.apple.com Hi Michael, Cheers, -- Jacques On Mar 1, 2008, at 4:24 PM, Michael Weiser wrote: Hello, When a program recursively creates directories, changes into them and calls getcwd(), it can create a rather deep directory structure before failing with error "No such file or directory": michael@esgaroth:~ # ./recgetcwd getcwd: No such file or directory .............................................................................................................................................................................................................................................................. michael@esgaroth:~ # sandbox-exec -f recgetcwd.sb ./recgetcwd mkdir: File name too long .............................................................................. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.a... You are right, it does sound like it may be a Seatbelt issue. Would you mind filing a report via < http://developer.apple.com/bugreporter/
? Among the benefits of filing it is that you can see the bug status and add additional information as needed. when trying to compile coreutils-6.10 in a sandbox, I noticed that their
getcwd() configure test yielded different results with and without the sandbox. I did some testing and boiled it down to the attached testcase: When running the same program in a sandbox that doesn't actualy restrict anything but has a rule containing a path name, it will fail much earlier with "File name too long": Has anyone come across this? -- bye, Micha <recgetcwd.c><recgetcwd.sb> _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/nectar %40apple.com This email sent to nectar@apple.com This email sent to site_archiver@lists.apple.com
participants (1)
-
Jacques Vidrine