#sydbox — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #sydbox, aggregated by home.social.
-
#Sydbox containers are not affected by the new LPE #Fragnesia because: 1. Unprivileged user/network namespaces are denied unless trace/allow_unsafe_namespace:user,net 2. Kernel algorithm (AF_ALG) sockets are denied unless trace/allow_unsafe_kcapi:true 3. Socket option TCP_ULP is denied unless trace/allow_unsafe_setsockopt:true. You may sleep in peace: https://raw.githubusercontent.com/v12-security/pocs/d4043edc2acbd75d093e3f5795751b678c66b259/fragnesia/fragnesia.c #exherbo #linux #security
-
#Sydbox containers are not affected by the new LPE #Fragnesia because: 1. Unprivileged user/network namespaces are denied unless trace/allow_unsafe_namespace:user,net 2. Kernel algorithm (AF_ALG) sockets are denied unless trace/allow_unsafe_kcapi:true 3. Socket option TCP_ULP is denied unless trace/allow_unsafe_setsockopt:true. You may sleep in peace: https://raw.githubusercontent.com/v12-security/pocs/d4043edc2acbd75d093e3f5795751b678c66b259/fragnesia/fragnesia.c #exherbo #linux #security
-
#Sydbox containers are not affected by the new LPE #Fragnesia because: 1. Unprivileged user/network namespaces are denied unless trace/allow_unsafe_namespace:user,net 2. Kernel algorithm (AF_ALG) sockets are denied unless trace/allow_unsafe_kcapi:true 3. Socket option TCP_ULP is denied unless trace/allow_unsafe_setsockopt:true. You may sleep in peace: https://raw.githubusercontent.com/v12-security/pocs/d4043edc2acbd75d093e3f5795751b678c66b259/fragnesia/fragnesia.c #exherbo #linux #security
-
#Sydbox containers are not affected by the new LPE #Fragnesia because: 1. Unprivileged user/network namespaces are denied unless trace/allow_unsafe_namespace:user,net 2. Kernel algorithm (AF_ALG) sockets are denied unless trace/allow_unsafe_kcapi:true 3. Socket option TCP_ULP is denied unless trace/allow_unsafe_setsockopt:true. You may sleep in peace: https://raw.githubusercontent.com/v12-security/pocs/d4043edc2acbd75d093e3f5795751b678c66b259/fragnesia/fragnesia.c #exherbo #linux #security
-
I made an #asciicast showcasing Ghost Mode of #sydbox: https://asciinema.org/a/1039185 #exherbo #linux #security
-
I made an #asciicast showcasing Ghost Mode of #sydbox: https://asciinema.org/a/1039185 #exherbo #linux #security
-
I made an #asciicast showcasing Ghost Mode of #sydbox: https://asciinema.org/a/1039185 #exherbo #linux #security
-
I made an #asciicast showcasing Ghost Mode of #sydbox: https://asciinema.org/a/1039185 #exherbo #linux #security
-
#Sydbox 3.53.0 is released! This is a feature release improving sandbox categories walk, stat, and adding the new category list for directory listing which allows easy use of walk+list categories for path hiding. readlink is also split from stat category which is by far the most common syscall so this helps with overhead of other categories. We also have bunch of security fixes. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md?ref_type=heads#3530 #exherbo #linux #security
-
#Sydbox 3.53.0 is released! This is a feature release improving sandbox categories walk, stat, and adding the new category list for directory listing which allows easy use of walk+list categories for path hiding. readlink is also split from stat category which is by far the most common syscall so this helps with overhead of other categories. We also have bunch of security fixes. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md?ref_type=heads#3530 #exherbo #linux #security
-
#Sydbox 3.53.0 is released! This is a feature release improving sandbox categories walk, stat, and adding the new category list for directory listing which allows easy use of walk+list categories for path hiding. readlink is also split from stat category which is by far the most common syscall so this helps with overhead of other categories. We also have bunch of security fixes. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md?ref_type=heads#3530 #exherbo #linux #security
-
#Sydbox 3.53.0 is released! This is a feature release improving sandbox categories walk, stat, and adding the new category list for directory listing which allows easy use of walk+list categories for path hiding. readlink is also split from stat category which is by far the most common syscall so this helps with overhead of other categories. We also have bunch of security fixes. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md?ref_type=heads#3530 #exherbo #linux #security
-
News from #Sydbox #git: New option trace/force_wx_open: Specify whether creating/writing open(2) family system calls for executables should be denied regardless of path. This option is restricted to creat, open, openat, and openat2 syscalls and may be combined with trace/force_umask option to confine filesystem as Write XOR Execute. New profile "wx" combines the new option with trace/force_umask:7177 to confine filesystem as W^X. User profile includes wx profile. #exherbo #linux #security
-
News from #Sydbox #git: New option trace/force_wx_open: Specify whether creating/writing open(2) family system calls for executables should be denied regardless of path. This option is restricted to creat, open, openat, and openat2 syscalls and may be combined with trace/force_umask option to confine filesystem as Write XOR Execute. New profile "wx" combines the new option with trace/force_umask:7177 to confine filesystem as W^X. User profile includes wx profile. #exherbo #linux #security
-
News from #Sydbox #git: New option trace/force_wx_open: Specify whether creating/writing open(2) family system calls for executables should be denied regardless of path. This option is restricted to creat, open, openat, and openat2 syscalls and may be combined with trace/force_umask option to confine filesystem as Write XOR Execute. New profile "wx" combines the new option with trace/force_umask:7177 to confine filesystem as W^X. User profile includes wx profile. #exherbo #linux #security
-
News from #Sydbox #git: New option trace/force_wx_open: Specify whether creating/writing open(2) family system calls for executables should be denied regardless of path. This option is restricted to creat, open, openat, and openat2 syscalls and may be combined with trace/force_umask option to confine filesystem as Write XOR Execute. New profile "wx" combines the new option with trace/force_umask:7177 to confine filesystem as W^X. User profile includes wx profile. #exherbo #linux #security
-
#Sydbox 3.52.0 is released! I've just merged 428 commits from next to main to make this release. It includes no new features, only bug fixes. Some of these bug fixes are security critical and you're recommended to upgrade as soon as possible. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md #exherbo #linux #security
-
#Sydbox 3.52.0 is released! I've just merged 428 commits from next to main to make this release. It includes no new features, only bug fixes. Some of these bug fixes are security critical and you're recommended to upgrade as soon as possible. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md #exherbo #linux #security
-
#Sydbox 3.52.0 is released! I've just merged 428 commits from next to main to make this release. It includes no new features, only bug fixes. Some of these bug fixes are security critical and you're recommended to upgrade as soon as possible. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md #exherbo #linux #security
-
#Sydbox 3.52.0 is released! I've just merged 428 commits from next to main to make this release. It includes no new features, only bug fixes. Some of these bug fixes are security critical and you're recommended to upgrade as soon as possible. Full story, as always, is in the ChangeLog, thanks for flying Syd: https://gitlab.exherbo.org/sydbox/sydbox/-/blob/main/ChangeLog.md #exherbo #linux #security
-
Mitigation against copy.fail in upcoming #Sydbox: Syd will reject to open SUID files regardless of mode unless the option trace/allow_unsafe_open_suid:1 is set. This does not prevent exploitation altogether as the attacker can write to files such as /etc/passwd, however it raises the bar with very little added cost. #exherbo #linux #security
-
Mitigation against copy.fail in upcoming #Sydbox: Syd will reject to open SUID files regardless of mode unless the option trace/allow_unsafe_open_suid:1 is set. This does not prevent exploitation altogether as the attacker can write to files such as /etc/passwd, however it raises the bar with very little added cost. #exherbo #linux #security
-
Mitigation against copy.fail in upcoming #Sydbox: Syd will reject to open SUID files regardless of mode unless the option trace/allow_unsafe_open_suid:1 is set. This does not prevent exploitation altogether as the attacker can write to files such as /etc/passwd, however it raises the bar with very little added cost. #exherbo #linux #security
-
Mitigation against copy.fail in upcoming #Sydbox: Syd will reject to open SUID files regardless of mode unless the option trace/allow_unsafe_open_suid:1 is set. This does not prevent exploitation altogether as the attacker can write to files such as /etc/passwd, however it raises the bar with very little added cost. #exherbo #linux #security
-
#GVisor supports only x86_64, arm64 yet they claim they run everywhere. #Sydbox passes tests on x86_64, i686, x32, arm64, armv7, ppc64, ppc64le, ppc, s390x, loongarch64, mips64el, and mipsel but I won't claim we are portable until we have mips64, mips, m68k and sparc! Huge thanks to Compile Farm people for enabling us to test Syd on various different architectures! #exherbo #linux #security
-
#GVisor supports only x86_64, arm64 yet they claim they run everywhere. #Sydbox passes tests on x86_64, i686, x32, arm64, armv7, ppc64, ppc64le, ppc, s390x, loongarch64, mips64el, and mipsel but I won't claim we are portable until we have mips64, mips, m68k and sparc! Huge thanks to Compile Farm people for enabling us to test Syd on various different architectures! #exherbo #linux #security
-
#GVisor supports only x86_64, arm64 yet they claim they run everywhere. #Sydbox passes tests on x86_64, i686, x32, arm64, armv7, ppc64, ppc64le, ppc, s390x, loongarch64, mips64el, and mipsel but I won't claim we are portable until we have mips64, mips, m68k and sparc! Huge thanks to Compile Farm people for enabling us to test Syd on various different architectures! #exherbo #linux #security
-
#GVisor supports only x86_64, arm64 yet they claim they run everywhere. #Sydbox passes tests on x86_64, i686, x32, arm64, armv7, ppc64, ppc64le, ppc, s390x, loongarch64, mips64el, and mipsel but I won't claim we are portable until we have mips64, mips, m68k and sparc! Huge thanks to Compile Farm people for enabling us to test Syd on various different architectures! #exherbo #linux #security
-
#GVisor supports only x86_64, arm64 yet they claim they run everywhere. #Sydbox passes tests on x86_64, i686, x32, arm64, armv7, ppc64, ppc64le, ppc, s390x, loongarch64, mips64el, and mipsel but I won't claim we are portable until we have mips64, mips, m68k and sparc! Huge thanks to Compile Farm people for enabling us to test Syd on various different architectures! #exherbo #linux #security
-
Correction: I was wrong about copy.fail and #sydbox earlier: Force sandboxing and Crypt sandboxing _imply_ the option trace/allow_safe_kcapi:1 so when these two are in use the sandbox process can abuse the AEAD issue in the #Linux #kernel. With #sydbox 3.52.0 to be released very soon, we rename the trace/allow_safe_kcapi option to trace/allow_unsafe_kcapi and Force/Crypt sandboxing are no longer going to imply this option, rather allow only Syd's use of AF_ALG sockets. #exherbo #linux #security
-
Correction: I was wrong about copy.fail and #sydbox earlier: Force sandboxing and Crypt sandboxing _imply_ the option trace/allow_safe_kcapi:1 so when these two are in use the sandbox process can abuse the AEAD issue in the #Linux #kernel. With #sydbox 3.52.0 to be released very soon, we rename the trace/allow_safe_kcapi option to trace/allow_unsafe_kcapi and Force/Crypt sandboxing are no longer going to imply this option, rather allow only Syd's use of AF_ALG sockets. #exherbo #linux #security
-
Correction: I was wrong about copy.fail and #sydbox earlier: Force sandboxing and Crypt sandboxing _imply_ the option trace/allow_safe_kcapi:1 so when these two are in use the sandbox process can abuse the AEAD issue in the #Linux #kernel. With #sydbox 3.52.0 to be released very soon, we rename the trace/allow_safe_kcapi option to trace/allow_unsafe_kcapi and Force/Crypt sandboxing are no longer going to imply this option, rather allow only Syd's use of AF_ALG sockets. #exherbo #linux #security
-
Correction: I was wrong about copy.fail and #sydbox earlier: Force sandboxing and Crypt sandboxing _imply_ the option trace/allow_safe_kcapi:1 so when these two are in use the sandbox process can abuse the AEAD issue in the #Linux #kernel. With #sydbox 3.52.0 to be released very soon, we rename the trace/allow_safe_kcapi option to trace/allow_unsafe_kcapi and Force/Crypt sandboxing are no longer going to imply this option, rather allow only Syd's use of AF_ALG sockets. #exherbo #linux #security
-
Correction: I was wrong about copy.fail and #sydbox earlier: Force sandboxing and Crypt sandboxing _imply_ the option trace/allow_safe_kcapi:1 so when these two are in use the sandbox process can abuse the AEAD issue in the #Linux #kernel. With #sydbox 3.52.0 to be released very soon, we rename the trace/allow_safe_kcapi option to trace/allow_unsafe_kcapi and Force/Crypt sandboxing are no longer going to imply this option, rather allow only Syd's use of AF_ALG sockets. #exherbo #linux #security
-
New #container breakout: https://copy.fail/ #sydbox containers aren't affected because Syd denies access to Kernel Cryptography API (KCAPI, AF_ALG sockets) by default unless trace/allow_safe_kcapi:1 is specified at startup. Crypt Sandboxing is also not affected because we don't use AEAD but CTR(AES). #exherbo #linux #security
-
New #container breakout: https://copy.fail/ #sydbox containers aren't affected because Syd denies access to Kernel Cryptography API (KCAPI, AF_ALG sockets) by default unless trace/allow_safe_kcapi:1 is specified at startup. Crypt Sandboxing is also not affected because we don't use AEAD but CTR(AES). #exherbo #linux #security
-
New #container breakout: https://copy.fail/ #sydbox containers aren't affected because Syd denies access to Kernel Cryptography API (KCAPI, AF_ALG sockets) by default unless trace/allow_safe_kcapi:1 is specified at startup. Crypt Sandboxing is also not affected because we don't use AEAD but CTR(AES). #exherbo #linux #security
-
New #container breakout: https://copy.fail/ #sydbox containers aren't affected because Syd denies access to Kernel Cryptography API (KCAPI, AF_ALG sockets) by default unless trace/allow_safe_kcapi:1 is specified at startup. Crypt Sandboxing is also not affected because we don't use AEAD but CTR(AES). #exherbo #linux #security
-
New #container breakout: https://copy.fail/ #sydbox containers aren't affected because Syd denies access to Kernel Cryptography API (KCAPI, AF_ALG sockets) by default unless trace/allow_safe_kcapi:1 is specified at startup. Crypt Sandboxing is also not affected because we don't use AEAD but CTR(AES). #exherbo #linux #security
-
Symbolic links bite again! This time it's #NixOS did you know #sydbox has trace/force_no_symlinks and trace/force_no_magiclinks options to disable following symlinks/magiclinks? You can even change them at runtime to achieve #pledge like confinement: https://discourse.nixos.org/t/nix-security-advisory-privilege-escalation-via-symlink-following-during-fod-output-registration/76900 #nix #linux #security
-
Symbolic links bite again! This time it's #NixOS did you know #sydbox has trace/force_no_symlinks and trace/force_no_magiclinks options to disable following symlinks/magiclinks? You can even change them at runtime to achieve #pledge like confinement: https://discourse.nixos.org/t/nix-security-advisory-privilege-escalation-via-symlink-following-during-fod-output-registration/76900 #nix #linux #security