summaryrefslogtreecommitdiff
path: root/security/selinux/ss
AgeCommit message (Collapse)AuthorFilesLines
2023-10-04selinux: simplify avtab_insert_node() prototypeJacob Satterfield1-13/+7
__hashtab_insert() in hashtab.h has a cleaner interface that allows the caller to specify the chain node location that the new node is being inserted into so that it can update the node that currently occupies it. Signed-off-by: Jacob Satterfield <jsatterfield.linux@gmail.com> Reviewed-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-09-13selinux: hweight optimization in avtab_read_itemJacob Satterfield1-5/+2
avtab_read_item() is a hot function called when reading each rule in a binary policydb. With the current Fedora policy and refpolicy, this function is called nearly 100,000 times per policy load. A single avtab node is only permitted to have a single specifier to describe the data it holds. As such, a check is performed to make sure only one specifier is set. Previously this was done via a for-loop. However, there is already an optimal function for finding the number of bits set (hamming weight) and on some architectures, dedicated instructions (popcount) which can be executed much more efficiently. Even when using -mcpu=generic on a x86-64 Fedora 38 VM, this commit results in a modest 2-4% speedup for policy loading due to a substantial reduction in the number of instructions executed. Signed-off-by: Jacob Satterfield <jsatterfield.linux@gmail.com> Reviewed-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-09-13selinux: improve role transition hashingChristian Göttsche1-1/+1
The number of buckets is calculated by performing a binary AND against the mask of the hash table, which is one less than its size (which is a power of two). This leads to all top bits being discarded, e.g. with the Reference Policy on Debian there exists 376 entries, leading to a size of 512, discarding the top 23 bits. Use jhash to improve the hash table utilization: # current roletr: 376 entries and 124/512 buckets used, longest chain length 8, sum of chain length^2 1496 # patch roletr: 376 entries and 266/512 buckets used, longest chain length 4, sum of chain length^2 646 Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Reviewed-by: Stephen Smalley <stephen.smalley.work@gmail.com> [PM: line wrap in the commit description] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-09-13selinux: simplify avtab slot calculationChristian Göttsche1-7/+1
Instead of dividing by 8 and then performing log2 by hand, use a more readable calculation. The behavior of rounddown_pow_of_two() for an input of 0 is undefined, so handle that case and small values manually to achieve the same results. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Reviewed-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-09-13selinux: print sum of chain lengths^2 for hash tablesChristian Göttsche4-3/+9
Print the sum of chain lengths squared as a metric for hash tables to provide more insights, similar to avtabs. While on it add a comma in the avtab message to improve readability of the output. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Reviewed-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-09-12selinux: Annotate struct sidtab_str_cache with __counted_byKees Cook1-1/+1
Prepare for the coming implementation by GCC and Clang of the __counted_by attribute. Flexible array members annotated with __counted_by can have their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family functions). As found with Coccinelle[1], add __counted_by for struct sidtab_str_cache. [1] https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci Cc: Paul Moore <paul@paul-moore.com> Cc: Stephen Smalley <stephen.smalley.work@gmail.com> Cc: Eric Paris <eparis@parisplace.org> Cc: Ondrej Mosnacek <omosnace@redhat.com> Cc: selinux@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Reviewed-by: "Gustavo A. R. Silva" <gustavoars@kernel.org> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-30Merge tag 'selinux-pr-20230829' of ↵Linus Torvalds20-152/+147
git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux Pull selinux updates from Paul Moore: "Thirty three SELinux patches, which is a pretty big number for us, but there isn't really anything scary in here; in fact we actually manage to remove 10 lines of code with this :) - Promote the SELinux DEBUG_HASHES macro to CONFIG_SECURITY_SELINUX_DEBUG The DEBUG_HASHES macro was a buried SELinux specific preprocessor debug macro that was a problem waiting to happen. Promoting the debug macro to a proper Kconfig setting should help both improve the visibility of the feature as well enable improved test coverage. We've moved some additional debug functions under the CONFIG_SECURITY_SELINUX_DEBUG flag and we may see more work in the future. - Emit a pr_notice() message if virtual memory is executable by default As this impacts the SELinux access control policy enforcement, if the system's configuration is such that virtual memory is executable by default we print a single line notice to the console. - Drop avtab_search() in favor of avtab_search_node() Both functions are nearly identical so we removed avtab_search() and converted the callers to avtab_search_node(). - Add some SELinux network auditing helpers The helpers not only reduce a small amount of code duplication, but they provide an opportunity to improve UDP flood performance slightly by delaying initialization of the audit data in some cases. - Convert GFP_ATOMIC allocators to GFP_KERNEL when reading SELinux policy There were two SELinux policy load helper functions that were allocating memory using GFP_ATOMIC, they have been converted to GFP_KERNEL. - Quiet a KMSAN warning in selinux_inet_conn_request() A one-line error path (re)set patch that resolves a KMSAN warning. It is important to note that this doesn't represent a real bug in the current code, but it quiets KMSAN and arguably hardens the code against future changes. - Cleanup the policy capability accessor functions This is a follow-up to the patch which reverted SELinux to using a global selinux_state pointer. This patch cleans up some artifacts of that change and turns each accessor into a one-line READ_ONCE() call into the policy capabilities array. - A number of patches from Christian Göttsche Christian submitted almost two-thirds of the patches in this pull request as he worked to harden the SELinux code against type differences, variable overflows, etc. - Support for separating early userspace from the kernel in policy, with a later revert We did have a patch that added a new userspace initial SID which would allow SELinux to distinguish between early user processes created before the initial policy load and the kernel itself. Unfortunately additional post-merge testing revealed a problematic interaction with an old SELinux userspace on an old version of Ubuntu so we've reverted the patch until we can resolve the compatibility issue. - Remove some outdated comments dealing with LSM hook registration When we removed the runtime disable functionality we forgot to remove some old comments discussing the importance of LSM hook registration ordering. - Minor administrative changes Stephen Smalley updated his email address and "debranded" SELinux from "NSA SELinux" to simply "SELinux". We've come a long way from the original NSA submission and I would consider SELinux a true community project at this point so removing the NSA branding just makes sense" * tag 'selinux-pr-20230829' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux: (33 commits) selinux: prevent KMSAN warning in selinux_inet_conn_request() selinux: use unsigned iterator in nlmsgtab code selinux: avoid implicit conversions in policydb code selinux: avoid implicit conversions in selinuxfs code selinux: make left shifts well defined selinux: update type for number of class permissions in services code selinux: avoid implicit conversions in avtab code selinux: revert SECINITSID_INIT support selinux: use GFP_KERNEL while reading binary policy selinux: update comment on selinux_hooks[] selinux: avoid implicit conversions in services code selinux: avoid implicit conversions in mls code selinux: use identical iterator type in hashtab_duplicate() selinux: move debug functions into debug configuration selinux: log about VM being executable by default selinux: fix a 0/NULL mistmatch in ad_net_init_from_iif() selinux: introduce SECURITY_SELINUX_DEBUG configuration selinux: introduce and use lsm_ad_net_init*() helpers selinux: update my email address selinux: add missing newlines in pr_err() statements ...
2023-08-18selinux: set next pointer before attaching to listChristian Göttsche1-1/+1
Set the next pointer in filename_trans_read_helper() before attaching the new node under construction to the list, otherwise garbage would be dereferenced on subsequent failure during cleanup in the out goto label. Cc: <stable@vger.kernel.org> Fixes: 430059024389 ("selinux: implement new format of filename transitions") Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-10selinux: avoid implicit conversions in policydb codeChristian Göttsche1-32/+37
Use the identical type for local variables, e.g. loop counters. Declare members of struct policydb_compat_info unsigned to consistently use unsigned iterators. They hold read-only non-negative numbers in the global variable policydb_compat. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-10selinux: make left shifts well definedChristian Göttsche1-6/+6
The loops upper bound represent the number of permissions used (for the current class or in general). The limit for this is 32, thus we might left shift of one less, 31. Shifting a base of 1 results in undefined behavior; use (u32)1 as base. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-10selinux: update type for number of class permissions in services codeChristian Göttsche2-2/+2
Security classes have only up to 32 permissions, hence using an u16 is sufficient (while improving padding in struct selinux_mapping). Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-10selinux: avoid implicit conversions in avtab codeChristian Göttsche1-12/+12
Return u32 from avtab_hash() instead of int, since the hashing is done on u32 and the result is used as an index on the hash array. Use the type of the limit in for loops. Avoid signed to unsigned conversion of multiplication result in avtab_hash_eval() and perform multiplication in destination type. Use unsigned loop iterator for index operations, to avoid sign extension. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-09selinux: revert SECINITSID_INIT supportPaul Moore1-27/+0
This commit reverts 5b0eea835d4e ("selinux: introduce an initial SID for early boot processes") as it was found to cause problems on distros with old SELinux userspace tools/libraries, specifically Ubuntu 16.04. Hopefully we will be able to re-add this functionality at a later date, but let's revert this for now to help ensure a stable and backwards compatible SELinux tree. Link: https://lore.kernel.org/selinux/87edkseqf8.fsf@mail.lhotse Acked-by: Ondrej Mosnacek <omosnace@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-08selinux: use GFP_KERNEL while reading binary policyChristian Göttsche1-5/+5
Use GFP_KERNEL instead of GFP_ATOMIC while reading a binary policy in sens_read() and cat_read(), similar to surrounding code. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-04selinux: avoid implicit conversions in services codeChristian Göttsche1-11/+12
Use u32 as the output parameter type in security_get_classes() and security_get_permissions(), based on the type of the symtab nprim member. Declare the read-only class string parameter of security_get_permissions() const. Avoid several implicit conversions by using the identical type for the destination. Use the type identical to the source for local variables. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: cleanup extra whitespace in subject] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-04selinux: avoid implicit conversions in mls codeChristian Göttsche1-4/+6
Use u32 for ebitmap bits and sensitivity levels, char for the default range of a class. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: description tweaks] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-08-04selinux: use identical iterator type in hashtab_duplicate()Christian Göttsche1-1/+2
Use the identical type u32 for the loop iterator. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: remove extra whitespace in subject] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-28selinux: move debug functions into debug configurationChristian Göttsche5-3/+20
avtab_hash_eval() and hashtab_stat() are only used in policydb.c when the configuration SECURITY_SELINUX_DEBUG is enabled. Move the function definitions under that configuration as well and provide empty definitions in case SECURITY_SELINUX_DEBUG is disabled, to avoid using #ifdef in the callers. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-20selinux: introduce SECURITY_SELINUX_DEBUG configurationChristian Göttsche1-4/+4
The policy database code contains several debug output statements related to hashtable utilization. Those are guarded by the macro DEBUG_HASHES, which is neither documented nor set anywhere. Introduce a new Kconfig configuration guarding this and potential other future debugging related code. Disable the setting by default. Suggested-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: fixed line lengths in the help text] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: update my email addressStephen Smalley19-19/+19
Update my email address; MAINTAINERS was updated some time ago. Signed-off-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: add missing newlines in pr_err() statementsChristian Göttsche1-3/+3
The kernel print statements do not append an implicit newline to format strings. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: subject line tweak] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: drop avtab_search()Christian Göttsche4-42/+8
avtab_search() shares the same logic with avtab_search_node(), except that it returns, if found, a pointer to the struct avtab_node member datum instead of the node itself. Since the member is an embedded struct, and not a pointer, the returned value of avtab_search() and avtab_search_node() will always in unison either be NULL or non-NULL. Drop avtab_search() and replace its calls by avtab_search_node() to deduplicate logic and adopt the only caller caring for the type of the returned value accordingly. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: fix implicit conversions in the symtabChristian Göttsche2-2/+2
hashtab_init() takes an u32 as size parameter type. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: subject line tweaks] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: use consistent type for AV rule specifierChristian Göttsche3-3/+3
The specifier for avtab keys is always supplied with a type of u16, either as a macro to security_compute_sid() or the member specified of the struct avtab_key. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: avoid avtab overflowsChristian Göttsche1-2/+2
Prevent inserting more than the supported U32_MAX number of entries. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-19selinux: check for multiplication overflow in put_entry()Christian Göttsche1-2/+5
The function is always inlined and most of the time both relevant arguments are compile time constants, allowing compilers to elide the check. Also the function is part of outputting the policy, which is not performance critical. Also convert the type of the third parameter into a size_t, since it should always be a non-negative number of elements. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-07-10selinux: introduce an initial SID for early boot processesOndrej Mosnacek1-0/+27
Currently, SELinux doesn't allow distinguishing between kernel threads and userspace processes that are started before the policy is first loaded - both get the label corresponding to the kernel SID. The only way a process that persists from early boot can get a meaningful label is by doing a voluntary dyntransition or re-executing itself. Reusing the kernel label for userspace processes is problematic for several reasons: 1. The kernel is considered to be a privileged domain and generally needs to have a wide range of permissions allowed to work correctly, which prevents the policy writer from effectively hardening against early boot processes that might remain running unintentionally after the policy is loaded (they represent a potential extra attack surface that should be mitigated). 2. Despite the kernel being treated as a privileged domain, the policy writer may want to impose certain special limitations on kernel threads that may conflict with the requirements of intentional early boot processes. For example, it is a good hardening practice to limit what executables the kernel can execute as usermode helpers and to confine the resulting usermode helper processes. However, a (legitimate) process surviving from early boot may need to execute a different set of executables. 3. As currently implemented, overlayfs remembers the security context of the process that created an overlayfs mount and uses it to bound subsequent operations on files using this context. If an overlayfs mount is created before the SELinux policy is loaded, these "mounter" checks are made against the kernel context, which may clash with restrictions on the kernel domain (see 2.). To resolve this, introduce a new initial SID (reusing the slot of the former "init" initial SID) that will be assigned to any userspace process started before the policy is first loaded. This is easy to do, as we can simply label any process that goes through the bprm_creds_for_exec LSM hook with the new init-SID instead of propagating the kernel SID from the parent. To provide backwards compatibility for existing policies that are unaware of this new semantic of the "init" initial SID, introduce a new policy capability "userspace_initial_context" and set the "init" SID to the same context as the "kernel" SID unless this capability is set by the policy. Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-06-06selinux: avoid bool as identifier nameChristian Göttsche2-5/+5
Avoid using the identifier `bool` to improve support with future C standards. C23 is about to make `bool` a predefined macro (see N2654). Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-23selinux: deprecated fs oconChristian Göttsche2-1/+5
The object context type `fs`, not to be confused with the well used object context type `fscon`, was introduced in the initial git commit 1da177e4c3f4 ("Linux-2.6.12-rc2") but never actually used since. The paper "A Security Policy Configuration for the Security-Enhanced Linux" [1] mentions it under `7.2 File System Contexts` but also states: Currently, this configuration is unused. The policy statement defining such object contexts is `fscon`, e.g.: fscon 2 3 gen_context(system_u:object_r:conA_t,s0) \ gen_context(system_u:object_r:conB_t,s0) It is not documented at selinuxproject.org or in the SELinux notebook and not supported by the Reference Policy buildsystem - the statement is not properly sorted - and thus not used in the Reference or Fedora Policy. Print a warning message at policy load for each such object context: SELinux: void and deprecated fs ocon 02:03 This topic was initially highlighted by Nicolas Iooss [2]. [1]: https://media.defense.gov/2021/Jul/29/2002815735/-1/-1/0/SELINUX-SECURITY-POLICY-CONFIGURATION-REPORT.PDF [2]: https://lore.kernel.org/selinux/CAJfZ7=mP2eJaq2BfO3y0VnwUJaY2cS2p=HZMN71z1pKjzaT0Eg@mail.gmail.com/ Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: tweaked deprecation comment, description line wrapping] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-18selinux: keep context struct members in syncChristian Göttsche1-0/+1
Commit 53f3517ae087 ("selinux: do not leave dangling pointer behind") reset the `str` field of the `context` struct in an OOM error branch. In this struct the fields `str` and `len` are coupled and should be kept in sync. Set the length to zero according to the string be set to NULL. Fixes: 53f3517ae087 ("selinux: do not leave dangling pointer behind") Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-08selinux: small cleanups in selinux_audit_rule_init()Paul Moore1-20/+18
A few small tweaks to selinux_audit_rule_init(): - Adjust how we use the @rc variable so we are not doing any extra work in the common/success case. - Related to the above, rework the 'out' jump label so that the success and error paths are different, simplifying both. - Cleanup some of the vertical whitespace while we are making the other changes. Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-08selinux: declare read-only data arrays constChristian Göttsche1-1/+1
The array of mount tokens in only used in match_opt_prefix() and never modified. The array of symtab names is never modified and only used in the DEBUG_HASHES configuration as output. The array of files for the SElinux filesystem sub-directory `ss` is similar to the other `struct tree_descr` usages only read from to construct the containing entries. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-08selinux: retain const qualifier on string literal in avtab_hash_eval()Christian Göttsche2-2/+2
The second parameter `tag` of avtab_hash_eval() is only used for printing. In policydb_index() it is called with a string literal: avtab_hash_eval(&p->te_avtab, "rules"); Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: slight formatting tweak in description] Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-08selinux: adjust typos in commentsChristian Göttsche1-1/+1
Found by codespell(1) Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-05-08selinux: do not leave dangling pointer behindChristian Göttsche1-0/+1
In case mls_context_cpy() fails due to OOM set the free'd pointer in context_cpy() to NULL to avoid it potentially being dereferenced or free'd again in future. Freeing a NULL pointer is well-defined and a hard NULL dereference crash is at least not exploitable and should give a workable stack trace. Fixes: 12b29f34558b ("selinux: support deferred mapping of contexts") Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2023-03-14selinux: stop passing selinux_state pointers and their offspringStephen Smalley2-206/+141
Linus observed that the pervasive passing of selinux_state pointers introduced by me in commit aa8e712cee93 ("selinux: wrap global selinux state") adds overhead and complexity without providing any benefit. The original idea was to pave the way for SELinux namespaces but those have not yet been implemented and there isn't currently a concrete plan to do so. Remove the passing of the selinux_state pointers, reverting to direct use of the single global selinux_state, and likewise remove passing of child pointers like the selinux_avc. The selinux_policy pointer remains as it is needed for atomic switching of policies. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Reported-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/oe-kbuild-all/202303101057.mZ3Gv5fK-lkp@intel.com/ Signed-off-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-12-13Merge tag 'selinux-pr-20221212' of ↵Linus Torvalds4-45/+50
git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux Pull selinux updates from Paul Moore: "Two SELinux patches: one increases the sleep time on deprecated functionality, and one removes the indirect calls in the sidtab context conversion code" * tag 'selinux-pr-20221212' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux: selinux: remove the sidtab context conversion indirect calls selinux: increase the deprecation sleep for checkreqprot and runtime disable
2022-11-09selinux: remove the sidtab context conversion indirect callsPaul Moore4-44/+45
The sidtab conversion code has support for multiple context conversion routines through the use of function pointers and indirect calls. However, the reality is that all current users rely on the same conversion routine: convert_context(). This patch does away with this extra complexity and replaces the indirect calls with direct function calls; allowing us to remove a layer of obfuscation and create cleaner, more maintainable code. Reviewed-by: Ondrej Mosnacek <omosnace@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-10-19selinux: enable use of both GFP_KERNEL and GFP_ATOMIC in convert_context()GONG, Ruiqi3-5/+6
The following warning was triggered on a hardware environment: SELinux: Converting 162 SID table entries... BUG: sleeping function called from invalid context at __might_sleep+0x60/0x74 0x0 in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: 5943, name: tar CPU: 7 PID: 5943 Comm: tar Tainted: P O 5.10.0 #1 Call trace: dump_backtrace+0x0/0x1c8 show_stack+0x18/0x28 dump_stack+0xe8/0x15c ___might_sleep+0x168/0x17c __might_sleep+0x60/0x74 __kmalloc_track_caller+0xa0/0x7dc kstrdup+0x54/0xac convert_context+0x48/0x2e4 sidtab_context_to_sid+0x1c4/0x36c security_context_to_sid_core+0x168/0x238 security_context_to_sid_default+0x14/0x24 inode_doinit_use_xattr+0x164/0x1e4 inode_doinit_with_dentry+0x1c0/0x488 selinux_d_instantiate+0x20/0x34 security_d_instantiate+0x70/0xbc d_splice_alias+0x4c/0x3c0 ext4_lookup+0x1d8/0x200 [ext4] __lookup_slow+0x12c/0x1e4 walk_component+0x100/0x200 path_lookupat+0x88/0x118 filename_lookup+0x98/0x130 user_path_at_empty+0x48/0x60 vfs_statx+0x84/0x140 vfs_fstatat+0x20/0x30 __se_sys_newfstatat+0x30/0x74 __arm64_sys_newfstatat+0x1c/0x2c el0_svc_common.constprop.0+0x100/0x184 do_el0_svc+0x1c/0x2c el0_svc+0x20/0x34 el0_sync_handler+0x80/0x17c el0_sync+0x13c/0x140 SELinux: Context system_u:object_r:pssp_rsyslog_log_t:s0:c0 is not valid (left unmapped). It was found that within a critical section of spin_lock_irqsave in sidtab_context_to_sid(), convert_context() (hooked by sidtab_convert_params.func) might cause the process to sleep via allocating memory with GFP_KERNEL, which is problematic. As Ondrej pointed out [1], convert_context()/sidtab_convert_params.func has another caller sidtab_convert_tree(), which is okay with GFP_KERNEL. Therefore, fix this problem by adding a gfp_t argument for convert_context()/sidtab_convert_params.func and pass GFP_KERNEL/_ATOMIC properly in individual callers. Cc: stable@vger.kernel.org Link: https://lore.kernel.org/all/20221018120111.1474581-1-gongruiqi1@huawei.com/ [1] Reported-by: Tan Ninghao <tanninghao1@huawei.com> Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance") Signed-off-by: GONG, Ruiqi <gongruiqi1@huawei.com> Reviewed-by: Ondrej Mosnacek <omosnace@redhat.com> [PM: wrap long BUG() output lines, tweak subject line] Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-08-31selinux: declare read-only parameters constChristian Göttsche4-29/+31
Declare ebitmap, mls_level and mls_context parameters const where they are only read from. This allows callers to supply pointers to const as arguments and increases readability. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-06-15selinux: Add boundary check in put_entry()Xiu Jianfeng1-0/+2
Just like next_entry(), boundary check is necessary to prevent memory out-of-bound access. Signed-off-by: Xiu Jianfeng <xiujianfeng@huawei.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-06-14selinux: fix memleak in security_read_state_kernel()Xiu Jianfeng1-1/+8
In this function, it directly returns the result of __security_read_policy without freeing the allocated memory in *data, cause memory leak issue, so free the memory if __security_read_policy failed. Signed-off-by: Xiu Jianfeng <xiujianfeng@huawei.com> [PM: subject line tweak] Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-05-24Merge tag 'selinux-pr-20220523' of ↵Linus Torvalds3-55/+48
git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux Pull selinux updates from Paul Moore: "We've got twelve patches queued for v5.19, with most being fairly minor. The highlights are below: - The checkreqprot and runtime disable knobs have been deprecated for some time with no active users that we can find. In an effort to move things along we are adding a pause when the knobs are used to help make the deprecation more noticeable in case anyone is still using these hacks in the shadows. - We've added the anonymous inode class name to the AVC audit records when anonymous inodes are involved. This should make writing policy easier when anonymous inodes are involved. - More constification work. This is fairly straightforward and the source of most of the diffstat. - The usual minor cleanups: remove unnecessary assignments, assorted style/checkpatch fixes, kdoc fixes, macro while-loop encapsulations, #include tweaks, etc" * tag 'selinux-pr-20220523' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux: security: declare member holding string literal const selinux: log anon inode class name selinux: declare data arrays const selinux: fix indentation level of mls_ops block selinux: include necessary headers in headers selinux: avoid extra semicolon selinux: update parameter documentation selinux: resolve checkpatch errors selinux: don't sleep when CONFIG_SECURITY_SELINUX_CHECKREQPROT_VALUE is true selinux: checkreqprot is deprecated, add some ssleep() discomfort selinux: runtime disable is deprecated, add some ssleep() discomfort selinux: Remove redundant assignments
2022-05-18selinux: fix bad cleanup on error in hashtab_duplicate()Ondrej Mosnacek1-1/+2
The code attempts to free the 'new' pointer using kmem_cache_free(), which is wrong because this function isn't responsible of freeing it. Instead, the function should free new->htable and clear the contents of *new (to prevent double-free). Cc: stable@vger.kernel.org Fixes: c7c556f1e81b ("selinux: refactor changing booleans") Reported-by: Wander Lairson Costa <wander@redhat.com> Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-05-03selinux: declare data arrays constChristian Göttsche3-20/+18
The arrays for the policy capability names, the initial sid identifiers and the class and permission names are not changed at runtime. Declare them const to avoid accidental modification. Do not override the classmap and the initial sid list in the build time script genheaders. Check flose(3) is successful in genheaders.c, otherwise the written data might be corrupted or incomplete. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: manual merge due to fuzz, minor style tweaks] Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-05-03selinux: fix indentation level of mls_ops blockChristian Göttsche1-21/+21
Add one level of indentation to the code block of the label mls_ops in constraint_expr_eval(), to adjust the trailing break; to the parent case: branch. Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-05-03selinux: avoid extra semicolonChristian Göttsche1-9/+9
Wrap macro into `do { } while (0)` to avoid Clang emitting warnings about extra semicolons. Similar to userspace commit https://github.com/SELinuxProject/selinux/commit/9d85aa60d12e468e7fd510c2b5475b5299b71622 Signed-off-by: Christian Göttsche <cgzones@googlemail.com> [PM: whitespace/indenting tweaks] Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-05-03selinux: resolve checkpatch errorsChristian Göttsche1-8/+4
Reported by checkpatch: security/selinux/nlmsgtab.c --------------------------- ERROR: that open brace { should be on the previous line #29: FILE: security/selinux/nlmsgtab.c:29: +static const struct nlmsg_perm nlmsg_route_perms[] = +{ ERROR: that open brace { should be on the previous line #97: FILE: security/selinux/nlmsgtab.c:97: +static const struct nlmsg_perm nlmsg_tcpdiag_perms[] = +{ ERROR: that open brace { should be on the previous line #105: FILE: security/selinux/nlmsgtab.c:105: +static const struct nlmsg_perm nlmsg_xfrm_perms[] = +{ ERROR: that open brace { should be on the previous line #134: FILE: security/selinux/nlmsgtab.c:134: +static const struct nlmsg_perm nlmsg_audit_perms[] = +{ security/selinux/ss/policydb.c ------------------------------ ERROR: that open brace { should be on the previous line #318: FILE: security/selinux/ss/policydb.c:318: +static int (*destroy_f[SYM_NUM]) (void *key, void *datum, void *datap) = +{ ERROR: that open brace { should be on the previous line #674: FILE: security/selinux/ss/policydb.c:674: +static int (*index_f[SYM_NUM]) (void *key, void *datum, void *datap) = +{ ERROR: that open brace { should be on the previous line #1643: FILE: security/selinux/ss/policydb.c:1643: +static int (*read_f[SYM_NUM]) (struct policydb *p, struct symtab *s, void *fp) = +{ ERROR: that open brace { should be on the previous line #3246: FILE: security/selinux/ss/policydb.c:3246: + void *datap) = +{ Signed-off-by: Christian Göttsche <cgzones@googlemail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-04-04selinux: Remove redundant assignmentsMichal Orzel1-1/+0
Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends. Reported by clang-tidy [deadcode.DeadStores] Signed-off-by: Michal Orzel <michalorzel.eng@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
2022-03-22Merge tag 'selinux-pr-20220321' of ↵Linus Torvalds8-21/+13
git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux Pull selinux updates from Paul Moore: "We've got a number of SELinux patches queued up, the highlights are: - Fixup the security_fs_context_parse_param() LSM hook so it executes all of the LSM hook implementations unless a serious error occurs. We also correct the SELinux hook implementation so that it returns zero on success. - In addition to a few SELinux mount option parsing fixes, we simplified the parsing by moving it earlier in the process. The logic was that it was unlikely an admin/user would use the new mount API and not have the policy loaded before passing the SELinux options. - Properly fixed the LSM/SELinux/SCTP hooks with the addition of the security_sctp_assoc_established() hook. This work was done in conjunction with the netdev folks and should complete the move of the SCTP labeling from the endpoints to the associations. - Fixed a variety of sparse warnings caused by changes in the "__rcu" markings of some core kernel structures. - Ensure we access the superblock's LSM security blob using the stacking-safe accessors. - Added the ability for the kernel to always allow FIOCLEX and FIONCLEX if the "ioctl_skip_cloexec" policy capability is specified. - Various constifications improvements, type casting improvements, additional return value checks, and dead code/parameter removal. - Documentation fixes" * tag 'selinux-pr-20220321' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux: (23 commits) selinux: shorten the policy capability enum names docs: fix 'make htmldocs' warning in SCTP.rst selinux: allow FIOCLEX and FIONCLEX with policy capability selinux: use correct type for context length selinux: drop return statement at end of void functions security: implement sctp_assoc_established hook in selinux security: add sctp_assoc_established hook selinux: parse contexts for mount options early selinux: various sparse fixes selinux: try to use preparsed sid before calling parse_sid() selinux: Fix selinux_sb_mnt_opts_compat() LSM: general protection fault in legacy_parse_param selinux: fix a type cast problem in cred_init_security() selinux: drop unused macro selinux: simplify cred_init_security selinux: do not discard const qualifier in cast selinux: drop unused parameter of avtab_insert_node selinux: drop cast to same type selinux: enclose macro arguments in parenthesis selinux: declare name parameter of hash_eval const ...
generated by cgit v1.2.3 (git 2.39.0) at 2024-10-02 08:52:02 +0300