From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
stable@vger.kernel.org, Alexei Starovoitov <ast@kernel.org>,
Thomas Gleixner <tglx@linutronix.de>,
Ben Hutchings <ben.hutchings@codethink.co.uk>
Subject: [PATCH 4.4 65/88] bpf: Prevent memory disambiguation attack
Date: Fri, 14 Dec 2018 13:00:39 +0100 [thread overview]
Message-ID: <20181214115707.681167481@linuxfoundation.org> (raw)
In-Reply-To: <20181214115702.151309521@linuxfoundation.org>
4.4-stable review patch. If anyone has any objections, please let me know.
------------------
From: Alexei Starovoitov <ast@kernel.org>
commit af86ca4e3088fe5eacf2f7e58c01fa68ca067672 upstream.
Detect code patterns where malicious 'speculative store bypass' can be used
and sanitize such patterns.
39: (bf) r3 = r10
40: (07) r3 += -216
41: (79) r8 = *(u64 *)(r7 +0) // slow read
42: (7a) *(u64 *)(r10 -72) = 0 // verifier inserts this instruction
43: (7b) *(u64 *)(r8 +0) = r3 // this store becomes slow due to r8
44: (79) r1 = *(u64 *)(r6 +0) // cpu speculatively executes this load
45: (71) r2 = *(u8 *)(r1 +0) // speculatively arbitrary 'load byte'
// is now sanitized
Above code after x86 JIT becomes:
e5: mov %rbp,%rdx
e8: add $0xffffffffffffff28,%rdx
ef: mov 0x0(%r13),%r14
f3: movq $0x0,-0x48(%rbp)
fb: mov %rdx,0x0(%r14)
ff: mov 0x0(%rbx),%rdi
103: movzbq 0x0(%rdi),%rsi
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[bwh: Backported to 4.4:
- Add verifier_env parameter to check_stack_write()
- Look up stack slot_types with state->stack_slot_type[] rather than
state->stack[].slot_type[]
- Drop bpf_verifier_env argument to verbose()
- Adjust filename, context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
kernel/bpf/verifier.c | 63 ++++++++++++++++++++++++++++++++++++++++++++++----
1 file changed, 59 insertions(+), 4 deletions(-)
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -191,6 +191,7 @@ struct bpf_insn_aux_data {
enum bpf_reg_type ptr_type; /* pointer type for load/store insns */
struct bpf_map *map_ptr; /* pointer for call insn into lookup_elem */
};
+ int sanitize_stack_off; /* stack slot to be cleared */
bool seen; /* this insn was processed by the verifier */
};
@@ -569,8 +570,9 @@ static bool is_spillable_regtype(enum bp
/* check_stack_read/write functions track spill/fill of registers,
* stack boundary and alignment are checked in check_mem_access()
*/
-static int check_stack_write(struct verifier_state *state, int off, int size,
- int value_regno)
+static int check_stack_write(struct verifier_env *env,
+ struct verifier_state *state, int off,
+ int size, int value_regno, int insn_idx)
{
int i, spi = (MAX_BPF_STACK + off) / BPF_REG_SIZE;
/* caller checked that off % size == 0 and -MAX_BPF_STACK <= off < 0,
@@ -589,8 +591,32 @@ static int check_stack_write(struct veri
/* save register state */
state->spilled_regs[spi] = state->regs[value_regno];
- for (i = 0; i < BPF_REG_SIZE; i++)
+ for (i = 0; i < BPF_REG_SIZE; i++) {
+ if (state->stack_slot_type[MAX_BPF_STACK + off + i] == STACK_MISC &&
+ !env->allow_ptr_leaks) {
+ int *poff = &env->insn_aux_data[insn_idx].sanitize_stack_off;
+ int soff = (-spi - 1) * BPF_REG_SIZE;
+
+ /* detected reuse of integer stack slot with a pointer
+ * which means either llvm is reusing stack slot or
+ * an attacker is trying to exploit CVE-2018-3639
+ * (speculative store bypass)
+ * Have to sanitize that slot with preemptive
+ * store of zero.
+ */
+ if (*poff && *poff != soff) {
+ /* disallow programs where single insn stores
+ * into two different stack slots, since verifier
+ * cannot sanitize them
+ */
+ verbose("insn %d cannot access two stack slots fp%d and fp%d",
+ insn_idx, *poff, soff);
+ return -EINVAL;
+ }
+ *poff = soff;
+ }
state->stack_slot_type[MAX_BPF_STACK + off + i] = STACK_SPILL;
+ }
} else {
/* regular write of data into stack */
state->spilled_regs[spi] = (struct reg_state) {};
@@ -746,7 +772,8 @@ static int check_mem_access(struct verif
verbose("attempt to corrupt spilled pointer on stack\n");
return -EACCES;
}
- err = check_stack_write(state, off, size, value_regno);
+ err = check_stack_write(env, state, off, size,
+ value_regno, insn_idx);
} else {
err = check_stack_read(state, off, size, value_regno);
}
@@ -2228,6 +2255,34 @@ static int convert_ctx_accesses(struct v
else
continue;
+ if (type == BPF_WRITE &&
+ env->insn_aux_data[i + delta].sanitize_stack_off) {
+ struct bpf_insn patch[] = {
+ /* Sanitize suspicious stack slot with zero.
+ * There are no memory dependencies for this store,
+ * since it's only using frame pointer and immediate
+ * constant of zero
+ */
+ BPF_ST_MEM(BPF_DW, BPF_REG_FP,
+ env->insn_aux_data[i + delta].sanitize_stack_off,
+ 0),
+ /* the original STX instruction will immediately
+ * overwrite the same stack slot with appropriate value
+ */
+ *insn,
+ };
+
+ cnt = ARRAY_SIZE(patch);
+ new_prog = bpf_patch_insn_data(env, i + delta, patch, cnt);
+ if (!new_prog)
+ return -ENOMEM;
+
+ delta += cnt - 1;
+ env->prog = new_prog;
+ insn = new_prog->insnsi + i + delta;
+ continue;
+ }
+
if (env->insn_aux_data[i + delta].ptr_type != PTR_TO_CTX)
continue;
next prev parent reply other threads:[~2018-12-14 12:15 UTC|newest]
Thread overview: 105+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-12-14 11:59 [PATCH 4.4 00/88] 4.4.168-stable review Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 01/88] ipv6: Check available headroom in ip6_xmit() even without options Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 02/88] net: 8139cp: fix a BUG triggered by changing mtu with network traffic Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 03/88] net: phy: dont allow __set_phy_supported to add unsupported modes Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 04/88] net: Prevent invalid access to skb->prev in __qdisc_drop_all Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 05/88] rtnetlink: ndo_dflt_fdb_dump() only work for ARPHRD_ETHER devices Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 06/88] tcp: fix NULL ref in tail loss probe Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 07/88] tun: forbid iface creation with rtnl ops Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 08/88] neighbour: Avoid writing before skb->head in neigh_hh_output() Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 09/88] ipv4: ipv6: netfilter: Adjust the frag mem limit when truesize changes Greg Kroah-Hartman
2018-12-16 9:57 ` jwiesner
2018-12-14 11:59 ` [PATCH 4.4 10/88] ARM: OMAP2+: prm44xx: Fix section annotation on omap44xx_prm_enable_io_wakeup Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 11/88] ARM: OMAP1: ams-delta: Fix possible use of uninitialized field Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 12/88] sysv: return err instead of 0 in __sysv_write_inode Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 13/88] s390/cpum_cf: Reject request for sampling in event initialization Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 14/88] hwmon: (ina2xx) Fix current value calculation Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 15/88] ASoC: dapm: Recalculate audio map forcely when card instantiated Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 16/88] hwmon: (w83795) temp4_type has writable permission Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 17/88] Btrfs: send, fix infinite loop due to directory rename dependencies Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 18/88] ASoC: omap-mcpdm: Add pm_qos handling to avoid under/overruns with CPU_IDLE Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 19/88] ASoC: omap-dmic: Add pm_qos handling to avoid overruns " Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 20/88] exportfs: do not read dentry after free Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 21/88] bpf: fix check of allowed specifiers in bpf_trace_printk Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 22/88] USB: omap_udc: use devm_request_irq() Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 23/88] USB: omap_udc: fix crashes on probe error and module removal Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 24/88] USB: omap_udc: fix omap_udc_start() on 15xx machines Greg Kroah-Hartman
2018-12-14 11:59 ` [PATCH 4.4 25/88] USB: omap_udc: fix USB gadget functionality on Palm Tungsten E Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 26/88] KVM: x86: fix empty-body warnings Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 27/88] net: thunderx: fix NULL pointer dereference in nic_remove Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 28/88] ixgbe: recognize 1000BaseLX SFP modules as 1Gbps Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 29/88] net: hisilicon: remove unexpected free_netdev Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 30/88] drm/ast: fixed reading monitor EDID not stable issue Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 31/88] xen: xlate_mmu: add missing header to fix W=1 warning Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 32/88] fscache: fix race between enablement and dropping of object Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 33/88] fscache, cachefiles: remove redundant variable cache Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 34/88] ocfs2: fix deadlock caused by ocfs2_defrag_extent() Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 35/88] hfs: do not free node before using Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 36/88] hfsplus: " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 37/88] debugobjects: avoid recursive calls with kmemleak Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 38/88] ocfs2: fix potential use after free Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 39/88] pstore: Convert console write to use ->write_buf Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 40/88] ALSA: pcm: remove SNDRV_PCM_IOCTL1_INFO internal command Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 41/88] KVM: nVMX: fix msr bitmaps to prevent L2 from accessing L0 x2APIC Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 42/88] KVM: nVMX: mark vmcs12 pages dirty on L2 exit Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 43/88] KVM: nVMX: Eliminate vmcs02 pool Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 44/88] KVM: VMX: introduce alloc_loaded_vmcs Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 45/88] KVM: VMX: make MSR bitmaps per-VCPU Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 46/88] KVM/x86: Add IBPB support Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 47/88] KVM/VMX: Emulate MSR_IA32_ARCH_CAPABILITIES Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 48/88] KVM/VMX: Allow direct access to MSR_IA32_SPEC_CTRL Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 49/88] KVM/SVM: " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 50/88] KVM/x86: Remove indirect MSR op calls from SPEC_CTRL Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 51/88] x86: reorganize SMAP handling in user space accesses Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 52/88] x86: fix SMAP in 32-bit environments Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 53/88] x86: Introduce __uaccess_begin_nospec() and uaccess_try_nospec Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 54/88] x86/usercopy: Replace open coded stac/clac with __uaccess_{begin, end} Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 55/88] x86/uaccess: Use __uaccess_begin_nospec() and uaccess_try_nospec Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 56/88] x86/bugs, KVM: Support the combination of guest and host IBRS Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 57/88] x86/KVM/VMX: Expose SPEC_CTRL Bit(2) to the guest Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 58/88] KVM: SVM: Move spec control call after restore of GS Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 59/88] x86/bugs, KVM: Extend speculation control for VIRT_SPEC_CTRL Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 60/88] x86/speculation: Use synthetic bits for IBRS/IBPB/STIBP Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 61/88] KVM: SVM: Implement VIRT_SPEC_CTRL support for SSBD Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 62/88] bpf: support 8-byte metafield access Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 63/88] bpf/verifier: Add spi variable to check_stack_write() Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 64/88] bpf/verifier: Pass instruction index to check_mem_access() and check_xadd() Greg Kroah-Hartman
2018-12-14 12:00 ` Greg Kroah-Hartman [this message]
2018-12-14 12:00 ` [PATCH 4.4 66/88] wil6210: missing length check in wmi_set_ie Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 67/88] posix-timers: Sanitize overrun handling Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 68/88] mm/hugetlb.c: dont call region_abort if region_chg fails Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 69/88] hugetlbfs: fix offset overflow in hugetlbfs mmap Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 70/88] hugetlbfs: check for pgoff value overflow Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 71/88] hugetlbfs: fix bug in pgoff overflow checking Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 72/88] swiotlb: clean up reporting Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 73/88] sr: pass down correctly sized SCSI sense buffer Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 74/88] mm: remove write/force parameters from __get_user_pages_locked() Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 75/88] mm: remove write/force parameters from __get_user_pages_unlocked() Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 76/88] mm: replace get_user_pages_unlocked() write/force parameters with gup_flags Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 77/88] mm: replace get_user_pages_locked() " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 78/88] mm: replace get_vaddr_frames() " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 79/88] mm: replace get_user_pages() " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 80/88] mm: replace __access_remote_vm() write parameter " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 81/88] mm: replace access_remote_vm() " Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 82/88] proc: dont use FOLL_FORCE for reading cmdline and environment Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 83/88] proc: do not access cmdline nor environ from file-backed areas Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 84/88] media: dvb-frontends: fix i2c access helpers for KASAN Greg Kroah-Hartman
2018-12-14 12:00 ` [PATCH 4.4 85/88] matroxfb: fix size of memcpy Greg Kroah-Hartman
2018-12-14 12:01 ` [PATCH 4.4 86/88] staging: speakup: Replace strncpy with memcpy Greg Kroah-Hartman
2018-12-14 12:01 ` [PATCH 4.4 87/88] rocker: fix rocker_tlv_put_* functions for KASAN Greg Kroah-Hartman
2018-12-14 12:01 ` [PATCH 4.4 88/88] selftests: Move networking/timestamping from Documentation Greg Kroah-Hartman
2018-12-14 15:39 ` [PATCH 4.4 00/88] 4.4.168-stable review Guenter Roeck
2018-12-14 17:33 ` kernelci.org bot
2018-12-14 20:12 ` shuah
2018-12-15 2:10 ` Guenter Roeck
2018-12-15 8:07 ` Greg Kroah-Hartman
2018-12-15 15:45 ` Guenter Roeck
2018-12-16 23:58 ` Ben Hutchings
2018-12-17 9:05 ` Greg Kroah-Hartman
2018-12-17 13:46 ` Guenter Roeck
2018-12-17 19:08 ` Greg Kroah-Hartman
2018-12-17 20:12 ` Guenter Roeck
2018-12-17 20:52 ` Greg Kroah-Hartman
2018-12-15 11:15 ` Harsh Shandilya
2018-12-17 9:06 ` Greg Kroah-Hartman
2018-12-15 16:44 ` Dan Rue
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20181214115707.681167481@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=ast@kernel.org \
--cc=ben.hutchings@codethink.co.uk \
--cc=linux-kernel@vger.kernel.org \
--cc=stable@vger.kernel.org \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).