user/sven/linux.git/kernel, branch v5.1.17

ftrace/x86: Remove possible deadlock between register_kprobe() and ftrace_run_update_code()

2019-07-10T07:52:31Z

commit d5b844a2cf507fc7642c9ae80a9d585db3065c28 upstream. The commit 9f255b632bf12c4dd7 ("module: Fix livepatch/ftrace module text permissions race") causes a possible deadlock between register_kprobe() and ftrace_run_update_code() when ftrace is using stop_machine(). The existing dependency chain (in reverse order) is: -> #1 (text_mutex){+.+.}: validate_chain.isra.21+0xb32/0xd70 __lock_acquire+0x4b8/0x928 lock_acquire+0x102/0x230 __mutex_lock+0x88/0x908 mutex_lock_nested+0x32/0x40 register_kprobe+0x254/0x658 init_kprobes+0x11a/0x168 do_one_initcall+0x70/0x318 kernel_init_freeable+0x456/0x508 kernel_init+0x22/0x150 ret_from_fork+0x30/0x34 kernel_thread_starter+0x0/0xc -> #0 (cpu_hotplug_lock.rw_sem){++++}: check_prev_add+0x90c/0xde0 validate_chain.isra.21+0xb32/0xd70 __lock_acquire+0x4b8/0x928 lock_acquire+0x102/0x230 cpus_read_lock+0x62/0xd0 stop_machine+0x2e/0x60 arch_ftrace_update_code+0x2e/0x40 ftrace_run_update_code+0x40/0xa0 ftrace_startup+0xb2/0x168 register_ftrace_function+0x64/0x88 klp_patch_object+0x1a2/0x290 klp_enable_patch+0x554/0x980 do_one_initcall+0x70/0x318 do_init_module+0x6e/0x250 load_module+0x1782/0x1990 __s390x_sys_finit_module+0xaa/0xf0 system_call+0xd8/0x2d0 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(text_mutex); lock(cpu_hotplug_lock.rw_sem); lock(text_mutex); lock(cpu_hotplug_lock.rw_sem); It is similar problem that has been solved by the commit 2d1e38f56622b9b ("kprobes: Cure hotplug lock ordering issues"). Many locks are involved. To be on the safe side, text_mutex must become a low level lock taken after cpu_hotplug_lock.rw_sem. This can't be achieved easily with the current ftrace design. For example, arm calls set_all_modules_text_rw() already in ftrace_arch_code_modify_prepare(), see arch/arm/kernel/ftrace.c. This functions is called: + outside stop_machine() from ftrace_run_update_code() + without stop_machine() from ftrace_module_enable() Fortunately, the problematic fix is needed only on x86_64. It is the only architecture that calls set_all_modules_text_rw() in ftrace path and supports livepatching at the same time. Therefore it is enough to move text_mutex handling from the generic kernel/trace/ftrace.c into arch/x86/kernel/ftrace.c: ftrace_arch_code_modify_prepare() ftrace_arch_code_modify_post_process() This patch basically reverts the ftrace part of the problematic commit 9f255b632bf12c4dd7 ("module: Fix livepatch/ftrace module text permissions race"). And provides x86_64 specific-fix. Some refactoring of the ftrace code will be needed when livepatching is implemented for arm or nds32. These architectures call set_all_modules_text_rw() and use stop_machine() at the same time. Link: http://lkml.kernel.org/r/20190627081334.12793-1-pmladek@suse.com Fixes: 9f255b632bf12c4dd7 ("module: Fix livepatch/ftrace module text permissions race") Acked-by: Thomas Gleixner Reported-by: Miroslav Benes Reviewed-by: Miroslav Benes Reviewed-by: Josh Poimboeuf Signed-off-by: Petr Mladek [ As reviewed by Miroslav Benes , removed return value of ftrace_run_update_code() as it is a void function. ] Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Greg Kroah-Hartman

tracing/snapshot: Resize spare buffer if size changed

2019-07-10T07:52:29Z

commit 46cc0b44428d0f0e81f11ea98217fc0edfbeab07 upstream. Current snapshot implementation swaps two ring_buffers even though their sizes are different from each other, that can cause an inconsistency between the contents of buffer_size_kb file and the current buffer size. For example: # cat buffer_size_kb 7 (expanded: 1408) # echo 1 > events/enable # grep bytes per_cpu/cpu0/stats bytes: 1441020 # echo 1 > snapshot // current:1408, spare:1408 # echo 123 > buffer_size_kb // current:123, spare:1408 # echo 1 > snapshot // current:1408, spare:123 # grep bytes per_cpu/cpu0/stats bytes: 1443700 # cat buffer_size_kb 123 // != current:1408 And also, a similar per-cpu case hits the following WARNING: Reproducer: # echo 1 > per_cpu/cpu0/snapshot # echo 123 > buffer_size_kb # echo 1 > per_cpu/cpu0/snapshot WARNING: WARNING: CPU: 0 PID: 1946 at kernel/trace/trace.c:1607 update_max_tr_single.part.0+0x2b8/0x380 Modules linked in: CPU: 0 PID: 1946 Comm: bash Not tainted 5.2.0-rc6 #20 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-2.fc30 04/01/2014 RIP: 0010:update_max_tr_single.part.0+0x2b8/0x380 Code: ff e8 dc da f9 ff 0f 0b e9 88 fe ff ff e8 d0 da f9 ff 44 89 ee bf f5 ff ff ff e8 33 dc f9 ff 41 83 fd f5 74 96 e8 b8 da f9 ff <0f> 0b eb 8d e8 af da f9 ff 0f 0b e9 bf fd ff ff e8 a3 da f9 ff 48 RSP: 0018:ffff888063e4fca0 EFLAGS: 00010093 RAX: ffff888066214380 RBX: ffffffff99850fe0 RCX: ffffffff964298a8 RDX: 0000000000000000 RSI: 00000000fffffff5 RDI: 0000000000000005 RBP: 1ffff1100c7c9f96 R08: ffff888066214380 R09: ffffed100c7c9f9b R10: ffffed100c7c9f9a R11: 0000000000000003 R12: 0000000000000000 R13: 00000000ffffffea R14: ffff888066214380 R15: ffffffff99851060 FS: 00007f9f8173c700(0000) GS:ffff88806d000000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000714dc0 CR3: 0000000066fa6000 CR4: 00000000000006f0 Call Trace: ? trace_array_printk_buf+0x140/0x140 ? __mutex_lock_slowpath+0x10/0x10 tracing_snapshot_write+0x4c8/0x7f0 ? trace_printk_init_buffers+0x60/0x60 ? selinux_file_permission+0x3b/0x540 ? tracer_preempt_off+0x38/0x506 ? trace_printk_init_buffers+0x60/0x60 __vfs_write+0x81/0x100 vfs_write+0x1e1/0x560 ksys_write+0x126/0x250 ? __ia32_sys_read+0xb0/0xb0 ? do_syscall_64+0x1f/0x390 do_syscall_64+0xc1/0x390 entry_SYSCALL_64_after_hwframe+0x49/0xbe This patch adds resize_buffer_duplicate_size() to check if there is a difference between current/spare buffer sizes and resize a spare buffer if necessary. Link: http://lkml.kernel.org/r/20190625012910.13109-1-devel@etsukata.com Cc: stable@vger.kernel.org Fixes: ad909e21bbe69 ("tracing: Add internal tracing_snapshot() functions") Signed-off-by: Eiichi Tsukata Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Greg Kroah-Hartman

ptrace: Fix ->ptracer_cred handling for PTRACE_TRACEME

2019-07-10T07:52:28Z

commit 6994eefb0053799d2e07cd140df6c2ea106c41ee upstream. Fix two issues: When called for PTRACE_TRACEME, ptrace_link() would obtain an RCU reference to the parent's objective credentials, then give that pointer to get_cred(). However, the object lifetime rules for things like struct cred do not permit unconditionally turning an RCU reference into a stable reference. PTRACE_TRACEME records the parent's credentials as if the parent was acting as the subject, but that's not the case. If a malicious unprivileged child uses PTRACE_TRACEME and the parent is privileged, and at a later point, the parent process becomes attacker-controlled (because it drops privileges and calls execve()), the attacker ends up with control over two processes with a privileged ptrace relationship, which can be abused to ptrace a suid binary and obtain root privileges. Fix both of these by always recording the credentials of the process that is requesting the creation of the ptrace relationship: current_cred() can't change under us, and current is the proper subject for access control. This change is theoretically userspace-visible, but I am not aware of any code that it will actually break. Fixes: 64b875f7ac8a ("ptrace: Capture the ptracer's creds not PT_PTRACE_CAP") Signed-off-by: Jann Horn Acked-by: Oleg Nesterov Cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds Signed-off-by: Greg Kroah-Hartman

ftrace: Fix NULL pointer dereference in free_ftrace_func_mapper()

2019-07-10T07:52:28Z

[ Upstream commit 04e03d9a616c19a47178eaca835358610e63a1dd ] The mapper may be NULL when called from register_ftrace_function_probe() with probe->data == NULL. This issue can be reproduced as follow (it may be covered by compiler optimization sometime): / # cat /sys/kernel/debug/tracing/set_ftrace_filter #### all functions enabled #### / # echo foo_bar:dump > /sys/kernel/debug/tracing/set_ftrace_filter [ 206.949100] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000 [ 206.952402] Mem abort info: [ 206.952819] ESR = 0x96000006 [ 206.955326] Exception class = DABT (current EL), IL = 32 bits [ 206.955844] SET = 0, FnV = 0 [ 206.956272] EA = 0, S1PTW = 0 [ 206.956652] Data abort info: [ 206.957320] ISV = 0, ISS = 0x00000006 [ 206.959271] CM = 0, WnR = 0 [ 206.959938] user pgtable: 4k pages, 48-bit VAs, pgdp=0000000419f3a000 [ 206.960483] [0000000000000000] pgd=0000000411a87003, pud=0000000411a83003, pmd=0000000000000000 [ 206.964953] Internal error: Oops: 96000006 [#1] SMP [ 206.971122] Dumping ftrace buffer: [ 206.973677] (ftrace buffer empty) [ 206.975258] Modules linked in: [ 206.976631] Process sh (pid: 281, stack limit = 0x(____ptrval____)) [ 206.978449] CPU: 10 PID: 281 Comm: sh Not tainted 5.2.0-rc1+ #17 [ 206.978955] Hardware name: linux,dummy-virt (DT) [ 206.979883] pstate: 60000005 (nZCv daif -PAN -UAO) [ 206.980499] pc : free_ftrace_func_mapper+0x2c/0x118 [ 206.980874] lr : ftrace_count_free+0x68/0x80 [ 206.982539] sp : ffff0000182f3ab0 [ 206.983102] x29: ffff0000182f3ab0 x28: ffff8003d0ec1700 [ 206.983632] x27: ffff000013054b40 x26: 0000000000000001 [ 206.984000] x25: ffff00001385f000 x24: 0000000000000000 [ 206.984394] x23: ffff000013453000 x22: ffff000013054000 [ 206.984775] x21: 0000000000000000 x20: ffff00001385fe28 [ 206.986575] x19: ffff000013872c30 x18: 0000000000000000 [ 206.987111] x17: 0000000000000000 x16: 0000000000000000 [ 206.987491] x15: ffffffffffffffb0 x14: 0000000000000000 [ 206.987850] x13: 000000000017430e x12: 0000000000000580 [ 206.988251] x11: 0000000000000000 x10: cccccccccccccccc [ 206.988740] x9 : 0000000000000000 x8 : ffff000013917550 [ 206.990198] x7 : ffff000012fac2e8 x6 : ffff000012fac000 [ 206.991008] x5 : ffff0000103da588 x4 : 0000000000000001 [ 206.991395] x3 : 0000000000000001 x2 : ffff000013872a28 [ 206.991771] x1 : 0000000000000000 x0 : 0000000000000000 [ 206.992557] Call trace: [ 206.993101] free_ftrace_func_mapper+0x2c/0x118 [ 206.994827] ftrace_count_free+0x68/0x80 [ 206.995238] release_probe+0xfc/0x1d0 [ 206.995555] register_ftrace_function_probe+0x4a8/0x868 [ 206.995923] ftrace_trace_probe_callback.isra.4+0xb8/0x180 [ 206.996330] ftrace_dump_callback+0x50/0x70 [ 206.996663] ftrace_regex_write.isra.29+0x290/0x3a8 [ 206.997157] ftrace_filter_write+0x44/0x60 [ 206.998971] __vfs_write+0x64/0xf0 [ 206.999285] vfs_write+0x14c/0x2f0 [ 206.999591] ksys_write+0xbc/0x1b0 [ 206.999888] __arm64_sys_write+0x3c/0x58 [ 207.000246] el0_svc_common.constprop.0+0x408/0x5f0 [ 207.000607] el0_svc_handler+0x144/0x1c8 [ 207.000916] el0_svc+0x8/0xc [ 207.003699] Code: aa0003f8 a9025bf5 aa0103f5 f946ea80 (f9400303) [ 207.008388] ---[ end trace 7b6d11b5f542bdf1 ]--- [ 207.010126] Kernel panic - not syncing: Fatal exception [ 207.011322] SMP: stopping secondary CPUs [ 207.013956] Dumping ftrace buffer: [ 207.014595] (ftrace buffer empty) [ 207.015632] Kernel Offset: disabled [ 207.017187] CPU features: 0x002,20006008 [ 207.017985] Memory Limit: none [ 207.019825] ---[ end Kernel panic - not syncing: Fatal exception ]--- Link: http://lkml.kernel.org/r/20190606031754.10798-1-liwei391@huawei.com Signed-off-by: Wei Li Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Sasha Levin

module: Fix livepatch/ftrace module text permissions race

2019-07-10T07:52:28Z

[ Upstream commit 9f255b632bf12c4dd7fc31caee89aa991ef75176 ] It's possible for livepatch and ftrace to be toggling a module's text permissions at the same time, resulting in the following panic: BUG: unable to handle page fault for address: ffffffffc005b1d9 #PF: supervisor write access in kernel mode #PF: error_code(0x0003) - permissions violation PGD 3ea0c067 P4D 3ea0c067 PUD 3ea0e067 PMD 3cc13067 PTE 3b8a1061 Oops: 0003 [#1] PREEMPT SMP PTI CPU: 1 PID: 453 Comm: insmod Tainted: G O K 5.2.0-rc1-a188339ca5 #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-20181126_142135-anatol 04/01/2014 RIP: 0010:apply_relocate_add+0xbe/0x14c Code: fa 0b 74 21 48 83 fa 18 74 38 48 83 fa 0a 75 40 eb 08 48 83 38 00 74 33 eb 53 83 38 00 75 4e 89 08 89 c8 eb 0a 83 38 00 75 43 <89> 08 48 63 c1 48 39 c8 74 2e eb 48 83 38 00 75 32 48 29 c1 89 08 RSP: 0018:ffffb223c00dbb10 EFLAGS: 00010246 RAX: ffffffffc005b1d9 RBX: 0000000000000000 RCX: ffffffff8b200060 RDX: 000000000000000b RSI: 0000004b0000000b RDI: ffff96bdfcd33000 RBP: ffffb223c00dbb38 R08: ffffffffc005d040 R09: ffffffffc005c1f0 R10: ffff96bdfcd33c40 R11: ffff96bdfcd33b80 R12: 0000000000000018 R13: ffffffffc005c1f0 R14: ffffffffc005e708 R15: ffffffff8b2fbc74 FS: 00007f5f447beba8(0000) GS:ffff96bdff900000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffc005b1d9 CR3: 000000003cedc002 CR4: 0000000000360ea0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: klp_init_object_loaded+0x10f/0x219 ? preempt_latency_start+0x21/0x57 klp_enable_patch+0x662/0x809 ? virt_to_head_page+0x3a/0x3c ? kfree+0x8c/0x126 patch_init+0x2ed/0x1000 [livepatch_test02] ? 0xffffffffc0060000 do_one_initcall+0x9f/0x1c5 ? kmem_cache_alloc_trace+0xc4/0xd4 ? do_init_module+0x27/0x210 do_init_module+0x5f/0x210 load_module+0x1c41/0x2290 ? fsnotify_path+0x3b/0x42 ? strstarts+0x2b/0x2b ? kernel_read+0x58/0x65 __do_sys_finit_module+0x9f/0xc3 ? __do_sys_finit_module+0x9f/0xc3 __x64_sys_finit_module+0x1a/0x1c do_syscall_64+0x52/0x61 entry_SYSCALL_64_after_hwframe+0x44/0xa9 The above panic occurs when loading two modules at the same time with ftrace enabled, where at least one of the modules is a livepatch module: CPU0 CPU1 klp_enable_patch() klp_init_object_loaded() module_disable_ro() ftrace_module_enable() ftrace_arch_code_modify_post_process() set_all_modules_text_ro() klp_write_object_relocations() apply_relocate_add() *patches read-only code* - BOOM A similar race exists when toggling ftrace while loading a livepatch module. Fix it by ensuring that the livepatch and ftrace code patching operations -- and their respective permissions changes -- are protected by the text_mutex. Link: http://lkml.kernel.org/r/ab43d56ab909469ac5d2520c5d944ad6d4abd476.1560474114.git.jpoimboe@redhat.com Reported-by: Johannes Erdfelt Fixes: 444d13ff10fb ("modules: add ro_after_init support") Acked-by: Jessica Yu Reviewed-by: Petr Mladek Reviewed-by: Miroslav Benes Signed-off-by: Josh Poimboeuf Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Sasha Levin

tracing: avoid build warning with HAVE_NOP_MCOUNT

2019-07-10T07:52:28Z

[ Upstream commit cbdaeaf050b730ea02e9ab4ff844ce54d85dbe1d ] Selecting HAVE_NOP_MCOUNT enables -mnop-mcount (if gcc supports it) and sets CC_USING_NOP_MCOUNT. Reuse __is_defined (which is suitable for testing CC_USING_* defines) to avoid conditional compilation and fix the following gcc 9 warning on s390: kernel/trace/ftrace.c:2514:1: warning: ‘ftrace_code_disable’ defined but not used [-Wunused-function] Link: http://lkml.kernel.org/r/patch.git-1a82d13f33ac.your-ad-here.call-01559732716-ext-6629@work.hours Fixes: 2f4df0017baed ("tracing: Add -mcount-nop option support") Signed-off-by: Vasily Gorbik Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Sasha Levin

cpuset: restore sanity to cpuset_cpus_allowed_fallback()

2019-07-10T07:52:27Z

[ Upstream commit d477f8c202d1f0d4791ab1263ca7657bbe5cf79e ] In the case that a process is constrained by taskset(1) (i.e. sched_setaffinity(2)) to a subset of available cpus, and all of those are subsequently offlined, the scheduler will set tsk->cpus_allowed to the current value of task_cs(tsk)->effective_cpus. This is done via a call to do_set_cpus_allowed() in the context of cpuset_cpus_allowed_fallback() made by the scheduler when this case is detected. This is the only call made to cpuset_cpus_allowed_fallback() in the latest mainline kernel. However, this is not sane behavior. I will demonstrate this on a system running the latest upstream kernel with the following initial configuration: # grep -i cpu /proc/$$/status Cpus_allowed: ffffffff,fffffff Cpus_allowed_list: 0-63 (Where cpus 32-63 are provided via smt.) If we limit our current shell process to cpu2 only and then offline it and reonline it: # taskset -p 4 $$ pid 2272's current affinity mask: ffffffffffffffff pid 2272's new affinity mask: 4 # echo off > /sys/devices/system/cpu/cpu2/online # dmesg | tail -3 [ 2195.866089] process 2272 (bash) no longer affine to cpu2 [ 2195.872700] IRQ 114: no longer affine to CPU2 [ 2195.879128] smpboot: CPU 2 is now offline # echo on > /sys/devices/system/cpu/cpu2/online # dmesg | tail -1 [ 2617.043572] smpboot: Booting Node 0 Processor 2 APIC 0x4 We see that our current process now has an affinity mask containing every cpu available on the system _except_ the one we originally constrained it to: # grep -i cpu /proc/$$/status Cpus_allowed: ffffffff,fffffffb Cpus_allowed_list: 0-1,3-63 This is not sane behavior, as the scheduler can now not only place the process on previously forbidden cpus, it can't even schedule it on the cpu it was originally constrained to! Other cases result in even more exotic affinity masks. Take for instance a process with an affinity mask containing only cpus provided by smt at the moment that smt is toggled, in a configuration such as the following: # taskset -p f000000000 $$ # grep -i cpu /proc/$$/status Cpus_allowed: 000000f0,00000000 Cpus_allowed_list: 36-39 A double toggle of smt results in the following behavior: # echo off > /sys/devices/system/cpu/smt/control # echo on > /sys/devices/system/cpu/smt/control # grep -i cpus /proc/$$/status Cpus_allowed: ffffff00,ffffffff Cpus_allowed_list: 0-31,40-63 This is even less sane than the previous case, as the new affinity mask excludes all smt-provided cpus with ids less than those that were previously in the affinity mask, as well as those that were actually in the mask. With this patch applied, both of these cases end in the following state: # grep -i cpu /proc/$$/status Cpus_allowed: ffffffff,ffffffff Cpus_allowed_list: 0-63 The original policy is discarded. Though not ideal, it is the simplest way to restore sanity to this fallback case without reinventing the cpuset wheel that rolls down the kernel just fine in cgroup v2. A user who wishes for the previous affinity mask to be restored in this fallback case can use that mechanism instead. This patch modifies scheduler behavior by instead resetting the mask to task_cs(tsk)->cpus_allowed by default, and cpu_possible mask in legacy mode. I tested the cases above on both modes. Note that the scheduler uses this fallback mechanism if and only if _every_ other valid avenue has been traveled, and it is the last resort before calling BUG(). Suggested-by: Waiman Long Suggested-by: Phil Auld Signed-off-by: Joel Savitz Acked-by: Phil Auld Acked-by: Waiman Long Acked-by: Peter Zijlstra (Intel) Signed-off-by: Tejun Heo Signed-off-by: Sasha Levin

signal: remove the wrong signal_pending() check in restore_user_sigmask()

2019-07-10T07:52:20Z

commit 97abc889ee296faf95ca0e978340fb7b942a3e32 upstream. This is the minimal fix for stable, I'll send cleanups later. Commit 854a6ed56839 ("signal: Add restore_user_sigmask()") introduced the visible change which breaks user-space: a signal temporary unblocked by set_user_sigmask() can be delivered even if the caller returns success or timeout. Change restore_user_sigmask() to accept the additional "interrupted" argument which should be used instead of signal_pending() check, and update the callers. Eric said: : For clarity. I don't think this is required by posix, or fundamentally to : remove the races in select. It is what linux has always done and we have : applications who care so I agree this fix is needed. : : Further in any case where the semantic change that this patch rolls back : (aka where allowing a signal to be delivered and the select like call to : complete) would be advantage we can do as well if not better by using : signalfd. : : Michael is there any chance we can get this guarantee of the linux : implementation of pselect and friends clearly documented. The guarantee : that if the system call completes successfully we are guaranteed that no : signal that is unblocked by using sigmask will be delivered? Link: http://lkml.kernel.org/r/20190604134117.GA29963@redhat.com Fixes: 854a6ed56839a40f6b5d02a2962f48841482eec4 ("signal: Add restore_user_sigmask()") Signed-off-by: Oleg Nesterov Reported-by: Eric Wong Tested-by: Eric Wong Acked-by: "Eric W. Biederman" Acked-by: Arnd Bergmann Acked-by: Deepa Dinamani Cc: Michael Kerrisk Cc: Jens Axboe Cc: Davidlohr Bueso Cc: Jason Baron Cc: Thomas Gleixner Cc: Al Viro Cc: David Laight Cc: [5.0+] Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds Signed-off-by: Greg Kroah-Hartman

bpf: fix unconnected udp hooks

2019-07-03T11:13:43Z

commit 983695fa676568fc0fe5ddd995c7267aabc24632 upstream. Intention of cgroup bind/connect/sendmsg BPF hooks is to act transparently to applications as also stated in original motivation in 7828f20e3779 ("Merge branch 'bpf-cgroup-bind-connect'"). When recently integrating the latter two hooks into Cilium to enable host based load-balancing with Kubernetes, I ran into the issue that pods couldn't start up as DNS got broken. Kubernetes typically sets up DNS as a service and is thus subject to load-balancing. Upon further debugging, it turns out that the cgroupv2 sendmsg BPF hooks API is currently insufficient and thus not usable as-is for standard applications shipped with most distros. To break down the issue we ran into with a simple example: # cat /etc/resolv.conf nameserver 147.75.207.207 nameserver 147.75.207.208 For the purpose of a simple test, we set up above IPs as service IPs and transparently redirect traffic to a different DNS backend server for that node: # cilium service list ID Frontend Backend 1 147.75.207.207:53 1 => 8.8.8.8:53 2 147.75.207.208:53 1 => 8.8.8.8:53 The attached BPF program is basically selecting one of the backends if the service IP/port matches on the cgroup hook. DNS breaks here, because the hooks are not transparent enough to applications which have built-in msg_name address checks: # nslookup 1.1.1.1 ;; reply from unexpected source: 8.8.8.8#53, expected 147.75.207.207#53 ;; reply from unexpected source: 8.8.8.8#53, expected 147.75.207.208#53 ;; reply from unexpected source: 8.8.8.8#53, expected 147.75.207.207#53 [...] ;; connection timed out; no servers could be reached # dig 1.1.1.1 ;; reply from unexpected source: 8.8.8.8#53, expected 147.75.207.207#53 ;; reply from unexpected source: 8.8.8.8#53, expected 147.75.207.208#53 ;; reply from unexpected source: 8.8.8.8#53, expected 147.75.207.207#53 [...] ; <<>> DiG 9.11.3-1ubuntu1.7-Ubuntu <<>> 1.1.1.1 ;; global options: +cmd ;; connection timed out; no servers could be reached For comparison, if none of the service IPs is used, and we tell nslookup to use 8.8.8.8 directly it works just fine, of course: # nslookup 1.1.1.1 8.8.8.8 1.1.1.1.in-addr.arpa name = one.one.one.one. In order to fix this and thus act more transparent to the application, this needs reverse translation on recvmsg() side. A minimal fix for this API is to add similar recvmsg() hooks behind the BPF cgroups static key such that the program can track state and replace the current sockaddr_in{,6} with the original service IP. From BPF side, this basically tracks the service tuple plus socket cookie in an LRU map where the reverse NAT can then be retrieved via map value as one example. Side-note: the BPF cgroups static key should be converted to a per-hook static key in future. Same example after this fix: # cilium service list ID Frontend Backend 1 147.75.207.207:53 1 => 8.8.8.8:53 2 147.75.207.208:53 1 => 8.8.8.8:53 Lookups work fine now: # nslookup 1.1.1.1 1.1.1.1.in-addr.arpa name = one.one.one.one. Authoritative answers can be found from: # dig 1.1.1.1 ; <<>> DiG 9.11.3-1ubuntu1.7-Ubuntu <<>> 1.1.1.1 ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 51550 ;; flags: qr rd ra ad; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 1 ;; OPT PSEUDOSECTION: ; EDNS: version: 0, flags:; udp: 512 ;; QUESTION SECTION: ;1.1.1.1. IN A ;; AUTHORITY SECTION: . 23426 IN SOA a.root-servers.net. nstld.verisign-grs.com. 2019052001 1800 900 604800 86400 ;; Query time: 17 msec ;; SERVER: 147.75.207.207#53(147.75.207.207) ;; WHEN: Tue May 21 12:59:38 UTC 2019 ;; MSG SIZE rcvd: 111 And from an actual packet level it shows that we're using the back end server when talking via 147.75.207.20{7,8} front end: # tcpdump -i any udp [...] 12:59:52.698732 IP foo.42011 > google-public-dns-a.google.com.domain: 18803+ PTR? 1.1.1.1.in-addr.arpa. (38) 12:59:52.698735 IP foo.42011 > google-public-dns-a.google.com.domain: 18803+ PTR? 1.1.1.1.in-addr.arpa. (38) 12:59:52.701208 IP google-public-dns-a.google.com.domain > foo.42011: 18803 1/0/0 PTR one.one.one.one. (67) 12:59:52.701208 IP google-public-dns-a.google.com.domain > foo.42011: 18803 1/0/0 PTR one.one.one.one. (67) [...] In order to be flexible and to have same semantics as in sendmsg BPF programs, we only allow return codes in [1,1] range. In the sendmsg case the program is called if msg->msg_name is present which can be the case in both, connected and unconnected UDP. The former only relies on the sockaddr_in{,6} passed via connect(2) if passed msg->msg_name was NULL. Therefore, on recvmsg side, we act in similar way to call into the BPF program whenever a non-NULL msg->msg_name was passed independent of sk->sk_state being TCP_ESTABLISHED or not. Note that for TCP case, the msg->msg_name is ignored in the regular recvmsg path and therefore not relevant. For the case of ip{,v6}_recv_error() paths, picked up via MSG_ERRQUEUE, the hook is not called. This is intentional as it aligns with the same semantics as in case of TCP cgroup BPF hooks right now. This might be better addressed in future through a different bpf_attach_type such that this case can be distinguished from the regular recvmsg paths, for example. Fixes: 1cedee13d25a ("bpf: Hooks for sys_sendmsg") Signed-off-by: Daniel Borkmann Acked-by: Andrey Ignatov Acked-by: Martin KaFai Lau Acked-by: Martynas Pumputis Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Signed-off-by: Greg Kroah-Hartman

bpf: fix nested bpf tracepoints with per-cpu data

2019-07-03T11:13:43Z

commit 9594dc3c7e71b9f52bee1d7852eb3d4e3aea9e99 upstream. BPF_PROG_TYPE_RAW_TRACEPOINTs can be executed nested on the same CPU, as they do not increment bpf_prog_active while executing. This enables three levels of nesting, to support - a kprobe or raw tp or perf event, - another one of the above that irq context happens to call, and - another one in nmi context (at most one of which may be a kprobe or perf event). Fixes: 20b9d7ac4852 ("bpf: avoid excessive stack usage for perf_sample_data") Signed-off-by: Matt Mullins Acked-by: Andrii Nakryiko Acked-by: Daniel Borkmann Signed-off-by: Alexei Starovoitov Signed-off-by: Greg Kroah-Hartman