HES2011 - Jon Oberheide and Dan Rosenberg - Stackjacking
5 likes853 views
This document discusses bypassing security protections provided by the grsecurity and PaX patchsets on Linux kernels. It begins with an introduction and agenda, then provides background on Linux kernel security issues over the past decade. The presentation notes that an arbitrary kernel write is a common exploitation primitive, but that this is insufficient to escalate privileges when protections like grsecurity/PaX are in place. It then introduces the concept of "stackjacking", where an attacker leverages kernel stack memory disclosures, which are common low severity vulnerabilities, along with an arbitrary kernel write to bypass grsecurity/PaX protections without needing to introduce new code or modify control flow.
HES2011 - Jon Oberheide and Dan Rosenberg - Stackjacking
- 2. Introduction ● Jon Oberheide ● Dan Rosenberg Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #2
- 3. Introduction In reference to the HES CFP: “I get excited every time I see a conference add requirements to their talk selection along the lines of 'exploitation presentations must be against grsecurity/PaX' -- but then there never ends up being any presentations of this kind.” – spender pratt Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #3
- 4. Agenda ● A review of Linux kernel security ● Exploitation vs. grsecurity/PaX ● Bypassing grsecurity/PaX Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #4
- 5. A decade of kernel security Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #5
- 6. A decade of kernel security Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #6
- 7. Upstream attitude ● Security is hard when upstream ignores the problems ● Linux still hasn't had its “security awakening” Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #7
- 8. How about last year? ● 142 CVE's assigned ● 30% worse than the previous worst year (2009) ● Based on public CVE requests, issues tracked at Red Hat Bugzilla, and Eugene's tagged git tree ● Missing dozens of non-CVE vulnerabilities (i.e. the “Dan Carpenter factor”) ● 61 (43%) discovered by six people ● Kees (4), Brad (3), Tavis (7), Vasiliy (4), Dan (37), Nelson (6) Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #8
- 9. Kernel vulns in 2010 ● 12 known exploits for local privilege escalation ● 13 remotely triggerable issues ● 33 potential privilege escalations Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #9
- 10. Breakdown by Target 2 31 76 33 Core Distro Exotic Red Hat Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #10
- 11. Breakdown by Impact 13 1 26 7 65 Bypass 30 DOS Info Priv Esc? Priv Esc Nothing Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #11
- 12. Interesting exploits of 2010 ● full-nelson.c ● Combined three vulns to get a NULL write ● half-nelson.c ● First Linux kernel stack overflow (not buffer overflow) exploit ● linux-rds-exploit.c ● Arbitrary write in RDS packet family ● i-CAN-haz-MODHARDEN.c ● SLUB overflow in CAN packet family ● american-sign-language.c ● Exploit payload written in ACPI's ASL/AML Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #12
- 13. Agenda ● A review of Linux kernel security ● Exploitation vs. grsecurity/PaX ● Bypassing grsecurity/PaX Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #13
- 14. Traditional Linux exploitation ● Perhaps most general exploitation primitive is an arbitrary kernel write ● Sometimes occurs naturally, other times can be constructed (e.g. overwriting pointers in an overflow to trigger a write) Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #14
- 15. Linux exploitation examples ● Writes to known addresses (IDT) ● Function pointer overwrites ● Redirecting control flow to userspace ● Influencing privesc-related kernel data (eg. credentials structures) ● Relying on kallsyms and other info Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #15
- 16. Overview of grsecurity/PaX ● grsecurity/PaX ● Third-party patchset to harden Linux userspace/kernel security ● Attempts to prevent ● Introduction/execution of arbitrary code ● Execution of existing code out of original order ● Execution of existing code in original order with arbitrary data Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #16
- 17. grsecurity/PaX hardening ● Kernel hardening features: ● KERNEXEC ● Prevent the introduction of new executable code ● UDEREF ● Prevent invalid userspace pointer dereferences ● HIDESYM ● Hide info that may be useful to an attacker (kallsyms, slabinfo, kernel address leaks, etc) ● MODHARDEN ● Prevent auto-loading of crappy unused packet families (CAN, RDS, econet, etc) Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #17
- 18. Agenda ● A review of Linux kernel security ● Exploitation vs. grsecurity/PaX ● Bypassing grsecurity/PaX Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #18
- 19. The main event ● A technique we call stackjacking ● Enables the bypass of common grsecurity/PaX configurations with common exploit primatives ● Independently discovered, collaboratively exploited, with slightly different techniques Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #19
- 20. Plan of attack! ??? STACK JACKING ROOT ??? OVERVIEW ??? ??? ??? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #20
- 21. Target kernel assumptions ● Hardened kernel with grsec/PaX ● Config level GRKERNSEC_HIGH ● KERNEXEC ● UDEREF ● HIDESYM ● MODHARDEN ● Etc... Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #21
- 22. Stronger target assumptions ● Let's make some extra assumptions ● We like a challenge, and these are assumptions that may possibly be obtainable now or in the future ● Stronger target assumptions ● Zero knowledge of kernel address space ● Fully randomized kernel text/data ● Cannot introduce new code into kernel address space ● Cannot modify kernel control flow (eg. data-only) Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #22
- 23. Attacker assumption #1 ● Assumption: arbitrary kmem write ● A common kernel exploitation primitive ● Examples: RDS, MCAST_MSFILTER ● Other vulns can be turned into writes, e.g. overflowing into a pointer that's written to ● Wut? ● “You mean I can't escalate privs with an arbitrary kernel memory write normally?” NOPE. Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #23
- 24. Arbitrary write into the abyss DARKNESS! 0xffffffff kernel 0xc0000000 (TASK_SIZE) user No clue where to write! 0x00000000 Exploitation is infeasible. Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #24
- 25. What's the secret sauce? ARBITRARY WRITE + ? = <3 Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #25
- 26. The Great Sandwich? ARBITRARY WRITE + = <3 msuiche? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #26
- 27. Nah, he's taken + = <3 Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #27
- 28. Need to know something ● One way: arbitrary kmem disclosure ● procfs (2005) ● sctp (2008) ● move_pages (2009) ● pktcdvd (2010) ● Just dump entire address space! ● But these are rare! ● And in many instances, mitigated by grsec/PaX Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #28
- 29. Something more common? ● How about a more common vuln? ● Hints... ● Widely considered to be a useless vulnerability ● Commonly assigned a CVSS score of 1.9 (low) ● 25+ such vulnerabilities reported in 2010 ● Often referred to as a Dan Rosenbug ● Can you guess it??? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #29
- 30. KSTACK MEM DISCLOSURE! ARBITRARY WRITE + KSTACK LEAK = <3 Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #30
- 31. How does kstack leak help? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #31
- 32. A bit about Linux kernel stacks ● Each userspace thread is high address allocated a kernel stack grows down ● Stores stack frames for kernel syscalls and other metadata 4k/8k stack unused ● Most commonly 8k, some distros use 4k ● THREAD_SIZE = 2*PAGE_SIZE = low address 2*4086 = 8192 Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #32
- 33. Kernel stack mem disclosures ● Kstack mem disclosures ● Leak of memory from the kernel stack to userspace ● Common cause ● Copying a struct on the kstack back to userspace with uninitialized fields ● Improper initialization/memset, forgetting member assignment, structure padding/holes ● A frequent occurrence, especially in compat Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #33
- 34. Kernel stack mem disclosures struct foo { uint32_t bar; uint32_t leak; kstack frame kstack frame uint32_t baz; foo.bar }; sensitive data sensitive data foo.leak syscall() { foo.baz struct foo; . foo.bar = 1; . foo.baz = 2; . copy_to_user(foo); } 1) process 2) kstack is reused 3) foo struct is copied to makes syscall on subsequent userspace, leaking 4 and leaves syscall and struct bytes of kstack through sensitive data overlaps with uninitialized foo.leak on kstack sensitive data member Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #34
- 35. Plan of attack! Arbitrary write STACK JACKING ROOT Kstack disclosure OVERVIEW ??? ??? ??? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #35
- 36. What's useful on the kstack? ● Leak data off kstack? ● Sensitive data left behind? Not really... ● Leak addresses off kstack? ● Sensitive addresses left behind? Maybe... ● Pointers to known structures could be exploited ● Too specific of an attack! ● Need something more general ● kstack disclosures differ widely in size/offsets Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #36
- 37. Kernel stack addresses ● How about a leaking an address that: ● Is stored on the stack; and ● Points to an address on the stack ● These are pretty common ● Eg. pointers to local stack vars, saved ebp, etc ● But what does this gain us? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #37
- 38. Kernel stack self-discovery ● If we can leak an pointer 0xcdef2000 to the kstack off the kstack, we can calculate the base 0xcdef1234 kstack frame address of the kstack 0xdeadbeef kstack_base = addr & ~(THREAD_SIZE – 1); 0xcdef1234 kstack_base = 0xcdef1234 & ~(8192 – 1) . . kstack_base = 0xcdef0000 0xcdef0000 . We call this kstack self-discovery Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #38
- 39. Effective kstack discovery ● Not all kstack disclosures are alike ● May only leak a few bytes, non-consecutive ● How do we effectively self-discover? ● Manual analysis ● Figure out where kstack leak overlaps addresses ● Automatic analysis ● libkstack Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #39
- 40. Manual kstack self-discovery ● Manual, offline analysis ● 1. prime stack with random syscall ● 2. leak bytes, see if any leaks match real kstack ● 3. repeat until we've collected enough bytes ● 4. construct list of priming syscalls needed for the particular leak to spill the beans Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #40
- 41. Automatic with libkstack ● We can automate this process for runtime self-discovery with libkstack ● 1. prime stack with random syscall ● 2. leak bytes, infer whether bytes belong to a kstack addr ● 3. repeat until we have sufficient confidence to calculate the kstack base addr Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #41
- 42. Plan of attack! Arbitrary write STACK JACKING ROOT Kstack disclosure OVERVIEW STACK SELF-DISCOVERY Manual analysis ??? ??? Auto with libkstack Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #42
- 43. No longer complete darkness A random pinpoint of light! 0xffffffff kernel 0xc0000000 (TASK_SIZE) kstack user We can self-discover kstack address! 0x00000000 Exploitation is...maybe feasible? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #43
- 44. The next step ● We now have a tiny island ● Use arbitrary write to modify anything on kstack ● Where to write? ● Pointers, data, metadata on kstack ● What to write? ● No userspace addrs (UDEREF), limited kernel ● Game over? Not yet! Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #44
- 45. Metadata on kernel stack Anything else of interest on the kstack??? high address start of stack grows down 4k/8k stack pointer stack unused thread_info current_thread_info low address thread_info struct stashed at base of kstack! Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #45
- 46. thread_info candidates struct thread_info { struct task_struct *task; struct exec_domain *exec_domain; __u32 flags; ● What can we __u32 status; __u32 cpu; modify within int preempt_count; mm_segment_t addr_limit; thread_info to struct restart_block restart_block; void __user *sysenter_return; escalate privs? #ifdef CONFIG_X86_32 unsigned long previous_esp; __u8 supervisor_stack; #endif int uaccess_err; }; Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #46
- 47. restart_block func ptr? struct thread_info { struct task_struct *task; struct exec_domain *exec_domain; ● restart_block? __u32 flags; __u32 status; ● Has a func ptr we __u32 cpu; int preempt_count; can overwrite and mm_segment_t addr_limit; invoke via userspace! struct restart_block restart_block; void __user *sysenter_return; ● Can't point to #ifdef CONFIG_X86_32 userspace (UDEREF) unsigned long previous_esp; __u8 supervisor_stack; ● Can't point to kmem #endif (blackbox) int uaccess_err; }; ● Plus assuming no control flow mod Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #47
- 48. task_struct pointer? struct thread_info { struct task_struct *task; struct exec_domain *exec_domain; ● task_struct? __u32 flags; __u32 status; ● Could point it at __u32 cpu; int preempt_count; init_task_struct for mm_segment_t addr_limit; getting creds/caps of struct restart_block restart_block; the init task void __user *sysenter_return; #ifdef CONFIG_X86_32 ● But we don't know unsigned long previous_esp; the address of __u8 supervisor_stack; #endif init_task_struct! int uaccess_err; }; Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #48
- 49. Attacking task_struct struct thread_info { struct task_struct *task; ... ● task_struct->creds? }; ● Modify creds of our process struct task_struct { ... directly to escalate privileges? const struct cred *real_cred; ● But in order to write const struct cred *cred; ... task_struct->creds, we need }; to know the address of task_struct! struct cred { ... ● If we could read the address uid_t uid; of task_struct off the end of gid_t gid; ... the kstack, we might win! }; Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #49
- 50. Connecting the dots Expanding our visibility 0xffffffff kernel 0xc0000000 creds (TASK_SIZE) task_struct kstack user If we can read off the kstack, 0x00000000 we can find task_struct/creds! Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #50
- 51. Attacking task_struct ● We have an arbitrary write on kstack ● Can we turn this into an arbitrary read? ● If we can get arbitrary read: ● Read base of kstack to find address of task_struct ● Read task_struct to find address of creds struct ● Write into creds struct to set uids/gids/caps ● Spawn a root shell! Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #51
- 52. Plan of attack! Arbitrary write STACK JACKING ROOT Kstack disclosure OVERVIEW STACK STACK SELF-DISCOVERY JACKING Manual analysis ??? ??? Read thread / task Auto with libkstack Overwrite creds Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #52
- 53. The Rosengrope Technique Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #53
- 54. Vanilla kernel ● No segmentation, user/kernel separation enforced by paging ● copy_*_user functions check user pointers against addr_limit (per-thread variable in thread_info struct) ● On vanilla, setting addr_limit to KERNEL_DS (ULONG_MAX) gives arbitrary read/write (all checks pass) Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #54
- 55. set_fs() ● Sometimes kernel wants to reuse code with kernel pointer arguments ● kernel_sendmsg, kernel_recvmsg, etc. ● Calls set_fs(KERNEL_DS) to set addr_limit and allow copy_*_user functions to copy kernel-to-kernel ● Careful to make sure no user-influenced pointers are used Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #55
- 56. PAX_UDEREF ● Strict user/kernel separation using segmentation ● Reload segment registers at kernel traps, used during copy operations ● Fault on invalid access Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #56
- 57. PAX_UDEREF and KERNEL_DS ● Use %gs register to keep track of segment for source/dest of copy ● set_fs(KERNEL_DS) sets addr_limit and reloads %gs register to contain __KERNEL_DS segment selector Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #57
- 58. No more easy root... ● Writing KERNEL_DS to addr_limit is no longer sufficient ● Access checks on pointers will pass, but we'll still fault in copy functions because of incorrect segment registers Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #58
- 59. But... ● %gs register is reloaded on context switch (necessary to keep track of thread state) ● Reloaded based on contents of addr_limit! Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #59
- 60. Using KERNEL_DS trick ● Write KERNEL_DS into addr_limit of current thread ● Loop on write(pipefd, addr, size) ● Eventually, thread will be scheduled out at right moment (before copy_from_user) ● When thread resumes, %gs register will be reloaded with __KERNEL_DS, and read target will be copied into pipe buffer (kernel-to-kernel copying) ● Restore addr_limit and read Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #60
- 61. Plan of attack! Arbitrary write STACK JACKING ROOT Kstack disclosure OVERVIEW STACK STACK STACK SELF-DISCOVERY GROPING JACKING Manual analysis KERNEL_DS trick ??? Read thread / task Auto with libkstack Overwrite creds Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #61
- 62. Pros and cons of KERNEL_DS ● KERNEL_DS ● Pros: clean, simple, generic method to obtain arbitrary read from write+kleak ● Cons: depends on knowing the location of addr_limit member of thread_info ● It's possible to move thread_info out of the kstack! ● Any alternatives? ● Let's get a bit crazier... Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #62
- 63. The Obergrope Technique Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #63
- 64. The Obergrope Technique Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #64
- 65. The Obergrope Technique Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #65
- 66. Attacking the kstack frames ● A different approach ● Don't attack the thread_info metadata on kstack ● Attack the kstack frames themselves! ● End goal is a read ● How to read data by writing a stack frame? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #66
- 67. Observations ● Lots of kernel codepaths copy data to userland, via copy_to_user(), put_user(), etc ● There may be copy_to_user() calls that use a source address argument that is, at some point, stored on the kernel stack ● If we can overwrite that source address on the kstack, we can control source of the copy_to_user() and leak data to userspace Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #67
- 68. A problem ● How can we write to our own kstack? ● Unlikely to be able to write into our own stack while exploiting the vulnerability for our arbitrary write ● Use parent/child processes ● Child self-discovers kstack addr ● Passes kstack addr to parent ● Parent writes into child while child is in syscall Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #68
- 69. More problems ● How can we write to stack reliably? ● We have a tricky race to win: ● Parent needs to write into child's kstack between when the copy_to_user() source register is pushed and popped from the kstack ● This is a very small race window... . Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #69
- 70. Winning Linux kernel races ● How to win Linux kernel races ● Get very lucky w/scheduling on SMP machine ● Cause a resource to be in contention (eg. locks) ● Cause kernel to page in from slow I/O device (sgrakkyu) ● Ehhh... ● We might hose the kernel if we lose the race ● Anything better? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #70
- 71. A twist on winning races ● This isn't a “standard” race though ● We can have child execute ANY codepath that performs copy_to_user() with a src arg on kstack ● Enter, sleepy syscalls! ● Syscalls that allow us to put process to sleep for an arbitrary amount of time ● nanosleep, wait, select, etc Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #71
- 72. Sleepy syscall conditions ● Any of these sleepy syscalls have our required conditions? ● Needs to: ● Push a register to the stack ● Go to sleep for an arbitrary amount of time ● Pop that register off the stack ● Use that register as the source for copy_to_user() Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #72
- 73. compat_sys_waitid asmlinkage long compat_sys_waitid(int which, compat_pid_t pid, struct compat_siginfo __user *uinfo, int options, struct compat_rusage __user *uru) { struct rusage ru; ... ret = sys_waitid(which, pid, (siginfo_t __user *)&info, uru ? (struct rusage __user *)&ru : NULL); ... ret = put_compat_rusage(&ru, uru); ... } int put_compat_rusage(const struct rusage *r, struct compat_rusage __user *ru) { if (!access_ok(VERIFY_WRITE, ru, sizeof(*ru)) || __put_user(r>ru_utime.tv_sec, &ru>ru_utime.tv_sec) || ... } Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #73
- 74. compat_sys_waitid disasm Dump of assembler code for function compat_sys_waitid: ... 0xffffffff810aba4e <+62>: lea 0x140(%rbp),%r14 1) compat_sys_waitid() stores address of ru in r14 ... 0xffffffff810aba8b <+123>: callq 0xffffffff81063b70 2) compat_sys_waitid() calls sys_waitid() <sys_waitid> ... 3) sys_waitid() calls do_wait() 0xffffffff810abaae <+158>: mov %r14,%rdi 0xffffffff810abab1 <+161>: callq 0xffffffff810aa700 4) do_wait() pushes r14 on kstack <put_compat_rusage> ... 5) do_wait() sleeps indefinitely Dump of assembler code for function sys_waitid: 6) we clobber the saved r14 reg on the kstack ... 0xffffffff81063bf9 <+137>: callq 0xffffffff810637e0 7) do_wait() wakes up <do_wait> 8) do_wait() pops r14 off the kstack ... 9) do_wait() returns Dump of assembler code for function do_wait: 10) sys_waitid() returns ... 0xffffffff810637e6 <+6>: push %r14 11) compat_sys_waitid() calls put_compat_rusage() ... PROCESS GOES TO SLEEP HERE 12) put_compat_rusage() uses clobbered source addr ... 0xffffffff810639fb <+539>: pop %r14 13) put_user() copies from source addr to userspace ... Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #74
- 75. compat_sys_waitid reliability ● Is this reliable across kernel versions? ● Yes, tested on: ● Lucid default build vmlinuz-2.6.32-24-generic ● Lucid custom build vmlinuz-2.6.32.26+drm33.12 ● Vanilla build vmlinuz-2.6.36.3 ● Vanilla build + grsec vmlinuz-2.6.36.3-grsec ● How about compilers? ● Across most gcc 4.x? Needs more investigation ● Potentially could runtime fingerprint compiler Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #75
- 76. High-level exploit flow 1. jacker forks/execs groper 8. helper wakes up from sleep 2. groper gets its own kstack addr 9. groper returns from waitid 3. groper passes kstack addr up to 10. groper leaks task_struct address jacker back to userspace 4. groper forks/execs helper 11. groper passes leaked address back up with jacker 5. helper goes to sleep for a bit 12. steps 4-11 are repeated to leak task/cred addresses 6. groper calls waitid on helper 13. jacker modifies groper's cred 7. jacker overwrites the required struct in-place offset on groper's stack 14. groper forks off a root shell Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #76
- 77. Plan of attack! Arbitrary write STACK JACKING ROOT Kstack disclosure OVERVIEW STACK STACK STACK SELF-DISCOVERY GROPING JACKING Manual analysis KERNEL_DS trick ??? Read thread / task Auto with libkstack Clobber saved reg Overwrite creds Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #77
- 78. Live demo! ● Exploit against live hardened system... Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #78
- 79. Defenses? ● Mitigate the exploitation vectors? ● Remove thread_info metadata from kstack ● RANDKSTACK ● Eliminate all kstack disclosures? ● Clear kstack between syscalls? ● Compiler/toolchain foo? ● ??? Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #79
- 80. Greetz ● #busticati ● $1$kk1q85Xp$Id.gAcJOg7uelf36VQwJQ/ ● Those who were already aware of this bypass vector ;-) Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #80
- 81. Q&A QUESTIONS? Jon Oberheide Dan Rosenberg jon@oberheide.org dan.j.rosenberg@gmail.com Duo Security Virtual Security Research Stackjacking Your Way to grsecurity/PaX Bypass – Jon Oberheide / Dan Rosenberg Slide #81