Implement an alternative CFI scheme that merges both the fine-grained
nature of kCFI but also takes full advantage of the coarse grained
hardware CFI as provided by IBT.
To contrast:
kCFI is a pure software CFI scheme and relies on being able to read
text -- specifically the instruction *before* the target symbol, and
does the hash validation *before* doing the call (otherwise control
flow is compromised already).
FineIBT is a software and hardware hybrid scheme; by ensuring every
branch target starts with a hash validation it is possible to place
the hash validation after the branch. This has several advantages:
o the (hash) load is avoided; no memop; no RX requirement.
o IBT WAIT-FOR-ENDBR state is a speculation stop; by placing
the hash validation in the immediate instruction after
the branch target there is a minimal speculation window
and the whole is a viable defence against SpectreBHB.
o Kees feels obliged to mention it is slightly more vulnerable
when the attacker can write code.
Obviously this patch relies on kCFI, but additionally it also relies
on the padding from the call-depth-tracking patches. It uses this
padding to place the hash-validation while the call-sites are
re-written to modify the indirect target to be 16 bytes in front of
the original target, thus hitting this new preamble.
Notably, there is no hardware that needs call-depth-tracking (Skylake)
and supports IBT (Tigerlake and onwards).
Suggested-by: Joao Moreira (Intel) <joao@overdrivepizza.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20221027092842.634714496@infradead.org
Add the location of all __cfi_##name symbols (as generated by kCFI) to
a section such that we might re-write things at kernel boot.
Notably; boot time re-hashing and FineIBT are the intended use of
this.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20221027092842.568039454@infradead.org
When code is compiled with:
-fpatchable-function-entry=${PADDING_BYTES},${PADDING_BYTES}
functions will have PADDING_BYTES of NOP in front of them. Unwinders
and other things that symbolize code locations will typically
attribute these bytes to the preceding function.
Given that these bytes nominally belong to the following symbol this
mis-attribution is confusing.
Inspired by the fact that CFI_CLANG emits __cfi_##name symbols to
claim these bytes, use objtool to emit __pfx_##name symbols to do
the same when CFI_CLANG is not used.
This then shows the callthunk for symbol 'name' as:
__pfx_##name+0x6/0x10
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Yujie Liu <yujie.liu@intel.com>
Link: https://lkml.kernel.org/r/20221028194453.592512209@infradead.org
When code is compiled with:
-fpatchable-function-entry=${PADDING_BYTES},${PADDING_BYTES}
functions will have PADDING_BYTES of NOP in front of them. Unwinders
and other things that symbolize code locations will typically
attribute these bytes to the preceding function.
Given that these bytes nominally belong to the following symbol this
mis-attribution is confusing.
Inspired by the fact that CFI_CLANG emits __cfi_##name symbols to
claim these bytes, allow objtool to emit __pfx_##name symbols to do
the same.
Therefore add the objtool --prefix=N argument, to conditionally place
a __pfx_##name symbol at N bytes ahead of symbol 'name' when: all
these preceding bytes are NOP and name-N is an instruction boundary.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Yujie Liu <yujie.liu@intel.com>
Link: https://lkml.kernel.org/r/20221028194453.526899822@infradead.org
Due to how gelf_update_sym*() requires an Elf_Data pointer, and how
libelf keeps Elf_Data in a linked list per section,
elf_update_symbol() ends up having to iterate this list on each
update to find the correct Elf_Data for the index'ed symbol.
By allocating one Elf_Data per new symbol, the list grows per new
symbol, giving an effective O(n^2) insertion time. This is obviously
bloody terrible.
Therefore over-allocate the Elf_Data when an extention is needed.
Except it turns out libelf disregards Elf_Scn::sh_size in favour of
the sum of Elf_Data::d_size. IOW it will happily write out all the
unused space and fill it with:
0000000000000000 0 NOTYPE LOCAL DEFAULT UND
entries (aka zeros). Which obviously violates the STB_LOCAL placement
rule, and is a general pain in the backside for not being the desired
behaviour.
Manually fix-up the Elf_Data size to avoid this problem before calling
elf_update().
This significantly improves performance when adding a significant
number of symbols.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Yujie Liu <yujie.liu@intel.com>
Link: https://lkml.kernel.org/r/20221028194453.461658986@infradead.org
In order to facilitate creation of more symbol types, slice up
elf_create_section_symbol() to extract a generic helper that deals
with adding ELF symbols.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Yujie Liu <yujie.liu@intel.com>
Link: https://lkml.kernel.org/r/20221028194453.396634875@infradead.org
This is a full revert of commit:
f138918162 ("kallsyms: Take callthunks into account")
The commit assumes a number of things that are not quite right.
Notably it assumes every symbol has PADDING_BYTES in front of it that
are not claimed by another symbol.
This is not true; even when compiled with:
-fpatchable-function-entry=${PADDING_BYTES},${PADDING_BYTES}
Notably things like .cold subfunctions do not need to adhere to this
change in ABI. It it also not true when build with CFI_CLANG, which
claims these PADDING_BYTES in the __cfi_##name symbol.
Once the prefix bytes are not consistent and or otherwise claimed the
approach this patch takes goes out the window and kallsym resolution
will report invalid symbol names.
Therefore revert this to make room for another approach.
Reported-by: Reported-by: kernel test robot <yujie.liu@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Yujie Liu <yujie.liu@intel.com>
Link: https://lore.kernel.org/r/202210241614.2ae4c1f5-yujie.liu@intel.com
Link: https://lkml.kernel.org/r/20221028194453.330970755@infradead.org
The first argument of WARN() is a condition, so this will use "addr"
as the format string and possibly crash.
Fixes: 3b6c1747da ("x86/retpoline: Add SKL retthunk retpolines")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/all/Y1gBoUZrRK5N%2FlCB@kili/
There's a conflict between the call-depth tracking commits in x86/core:
ee3e2469b3 ("x86/ftrace: Make it call depth tracking aware")
36b64f1012 ("x86/ftrace: Rebalance RSB")
eac828eaef ("x86/ftrace: Remove ftrace_epilogue()")
And these fixes in x86/urgent:
883bbbffa5 ("ftrace,kcfi: Separate ftrace_stub() and ftrace_stub_graph()")
b5f1fc3184 ("x86/ftrace: Remove ftrace_epilogue()")
It's non-trivial overlapping modifications - resolve them.
Conflicts:
arch/x86/kernel/ftrace_64.S
Signed-off-by: Ingo Molnar <mingo@kernel.org>
When an extended state component is not present in fpstate, but in init
state, the function copies from init_fpstate via copy_feature().
But, dynamic states are not present in init_fpstate because of all-zeros
init states. Then retrieving them from init_fpstate will explode like this:
BUG: kernel NULL pointer dereference, address: 0000000000000000
...
RIP: 0010:memcpy_erms+0x6/0x10
? __copy_xstate_to_uabi_buf+0x381/0x870
fpu_copy_guest_fpstate_to_uabi+0x28/0x80
kvm_arch_vcpu_ioctl+0x14c/0x1460 [kvm]
? __this_cpu_preempt_check+0x13/0x20
? vmx_vcpu_put+0x2e/0x260 [kvm_intel]
kvm_vcpu_ioctl+0xea/0x6b0 [kvm]
? kvm_vcpu_ioctl+0xea/0x6b0 [kvm]
? __fget_light+0xd4/0x130
__x64_sys_ioctl+0xe3/0x910
? debug_smp_processor_id+0x17/0x20
? fpregs_assert_state_consistent+0x27/0x50
do_syscall_64+0x3f/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd
Adjust the 'mask' to zero out the userspace buffer for the features that
are not available both from fpstate and from init_fpstate.
The dynamic features depend on the compacted XSAVE format. Ensure it is
enabled before reading XCOMP_BV in init_fpstate.
Fixes: 2308ee57d9 ("x86/fpu/amx: Enable the AMX feature in 64-bit mode")
Reported-by: Yuan Yao <yuan.yao@intel.com>
Suggested-by: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Tested-by: Yuan Yao <yuan.yao@intel.com>
Link: https://lore.kernel.org/lkml/BYAPR11MB3717EDEF2351C958F2C86EED95259@BYAPR11MB3717.namprd11.prod.outlook.com/
Link: https://lkml.kernel.org/r/20221021185844.13472-1-chang.seok.bae@intel.com
clear_cpu_cap(&boot_cpu_data) is very similar to setup_clear_cpu_cap()
except that the latter also sets a bit in 'cpu_caps_cleared' which
later clears the same cap in secondary cpus, which is likely what is
meant here.
Fixes: 47125db27e ("perf/x86/intel/lbr: Support Architectural LBR")
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Link: https://lkml.kernel.org/r/20220718141123.136106-2-mlevitsk@redhat.com
Different function signatures means they needs to be different
functions; otherwise CFI gets upset.
As triggered by the ftrace boot tests:
[] CFI failure at ftrace_return_to_handler+0xac/0x16c (target: ftrace_stub+0x0/0x14; expected type: 0x0a5d5347)
Fixes: 3c516f89e1 ("x86: Add support for CONFIG_CFI_CLANG")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Link: https://lkml.kernel.org/r/Y06dg4e1xF6JTdQq@hirez.programming.kicks-ass.net
Remove the weird jumps to RET and simply use RET.
This then promotes ftrace_stub() to a real function; which becomes
important for kcfi.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.719080593@infradead.org
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Fix
./include/trace/events/xen.h:28:31: warning: ‘enum paravirt_lazy_mode’ \
declared inside parameter list will not be visible outside of this definition or declaration
which turns into a build error:
./include/trace/events/xen.h:28:50: error: parameter 1 (‘mode’) has incomplete type
28 | TP_PROTO(enum paravirt_lazy_mode mode), \
due to enum paravirt_lazy_mode being visible only under CONFIG_PARAVIRT.
Just pull it up where it is unconditionally visible.
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/Y1AtAXM8YjtBm2cj@zn.tnic
AMD systems support zero CBM (capacity bit mask) for cache allocation.
That is reflected in rdt_init_res_defs_amd() by:
r->cache.arch_has_empty_bitmaps = true;
However given the unified code in cbm_validate(), checking for:
val == 0 && !arch_has_empty_bitmaps
is not enough because of another check in cbm_validate():
if ((zero_bit - first_bit) < r->cache.min_cbm_bits)
The default value of r->cache.min_cbm_bits = 1.
Leading to:
$ cd /sys/fs/resctrl
$ mkdir foo
$ cd foo
$ echo L3:0=0 > schemata
-bash: echo: write error: Invalid argument
$ cat /sys/fs/resctrl/info/last_cmd_status
Need at least 1 bits in the mask
Initialize the min_cbm_bits to 0 for AMD. Also, remove the default
setting of min_cbm_bits and initialize it separately.
After the fix:
$ cd /sys/fs/resctrl
$ mkdir foo
$ cd foo
$ echo L3:0=0 > schemata
$ cat /sys/fs/resctrl/info/last_cmd_status
ok
Fixes: 316e7f901f ("x86/resctrl: Add struct rdt_cache::arch_has_{sparse, empty}_bitmaps")
Co-developed-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Babu Moger <babu.moger@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: James Morse <james.morse@arm.com>
Reviewed-by: Reinette Chatre <reinette.chatre@intel.com>
Reviewed-by: Fenghua Yu <fenghua.yu@intel.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/lkml/20220517001234.3137157-1-eranian@google.com
Currently, the patch application logic checks whether the revision
needs to be applied on each logical CPU (SMT thread). Therefore, on SMT
designs where the microcode engine is shared between the two threads,
the application happens only on one of them as that is enough to update
the shared microcode engine.
However, there are microcode patches which do per-thread modification,
see Link tag below.
Therefore, drop the revision check and try applying on each thread. This
is what the BIOS does too so this method is very much tested.
Btw, change only the early paths. On the late loading paths, there's no
point in doing per-thread modification because if is it some case like
in the bugzilla below - removing a CPUID flag - the kernel cannot go and
un-use features it has detected are there early. For that, one should
use early loading anyway.
[ bp: Fixes does not contain the oldest commit which did check for
equality but that is good enough. ]
Fixes: 8801b3fcb5 ("x86/microcode/AMD: Rework container parsing")
Reported-by: Ștefan Talpalaru <stefantalpalaru@yahoo.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Tested-by: Ștefan Talpalaru <stefantalpalaru@yahoo.com>
Cc: <stable@vger.kernel.org>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216211
Today, core ID is assumed to be unique within each package.
But an AlderLake-N platform adds a Module level between core and package,
Linux excludes the unknown modules bits from the core ID, resulting in
duplicate core ID's.
To keep core ID unique within a package, Linux must include all APIC-ID
bits for known or unknown levels above the core and below the package
in the core ID.
It is important to understand that core ID's have always come directly
from the APIC-ID encoding, which comes from the BIOS. Thus there is no
guarantee that they start at 0, or that they are contiguous.
As such, naively using them for array indexes can be problematic.
[ dhansen: un-known -> unknown ]
Fixes: 7745f03eb3 ("x86/topology: Add CPUID.1F multi-die/package support")
Suggested-by: Len Brown <len.brown@intel.com>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Len Brown <len.brown@intel.com>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20221014090147.1836-5-rui.zhang@intel.com
CPUID.1F/B does not enumerate Package level explicitly, instead, all the
APIC-ID bits above the enumerated levels are assumed to be package ID
bits.
Current code gets package ID by shifting out all the APIC-ID bits that
Linux supports, rather than shifting out all the APIC-ID bits that
CPUID.1F enumerates. This introduces problems when CPUID.1F enumerates a
level that Linux does not support.
For example, on a single package AlderLake-N, there are 2 Ecore Modules
with 4 atom cores in each module. Linux does not support the Module
level and interprets the Module ID bits as package ID and erroneously
reports a multi module system as a multi-package system.
Fix this by using APIC-ID bits above all the CPUID.1F enumerated levels
as package ID.
[ dhansen: spelling fix ]
Fixes: 7745f03eb3 ("x86/topology: Add CPUID.1F multi-die/package support")
Suggested-by: Len Brown <len.brown@intel.com>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Len Brown <len.brown@intel.com>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20221014090147.1836-4-rui.zhang@intel.com
The coretemp driver supports up to a hard-coded limit of 128 cores.
Today, the driver can not support a core with an ID above that limit.
Yet, the encoding of core ID's is arbitrary (BIOS APIC-ID) and so they
may be sparse and they may be large.
Update the driver to map arbitrary core ID numbers into appropriate
array indexes so that 128 cores can be supported, no matter the encoding
of core ID's.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Acked-by: Len Brown <len.brown@intel.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20221014090147.1836-3-rui.zhang@intel.com
A recent change in LLVM made CONFIG_EFI_STUB unselectable because it no
longer pretends to support -mabi=ms, breaking the dependency in
Kconfig. Lack of CONFIG_EFI_STUB can prevent kernels from booting via
EFI in certain circumstances.
This check was added by
8f24f8c2fc ("efi/libstub: Annotate firmware routines as __efiapi")
to ensure that __attribute__((ms_abi)) was available, as -mabi=ms is
not actually used in any cflags.
According to the GCC documentation, this attribute has been supported
since GCC 4.4.7. The kernel currently requires GCC 5.1 so this check is
not necessary; even when that change landed in 5.6, the kernel required
GCC 4.9 so it was unnecessary then as well.
Clang supports __attribute__((ms_abi)) for all versions that are
supported for building the kernel so no additional check is needed.
Remove the 'depends on' line altogether to allow CONFIG_EFI_STUB to be
selected when CONFIG_EFI is enabled, regardless of compiler.
Fixes: 8f24f8c2fc ("efi/libstub: Annotate firmware routines as __efiapi")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Cc: stable@vger.kernel.org
Link: d1ad006a8f
The fully secure mitigation for RSB underflow on Intel SKL CPUs is IBRS,
which inflicts up to 30% penalty for pathological syscall heavy work loads.
Software based call depth tracking and RSB refill is not perfect, but
reduces the attack surface massively. The penalty for the pathological case
is about 8% which is still annoying but definitely more palatable than IBRS.
Add a retbleed=stuff command line option to enable the call depth tracking
and software refill of the RSB.
This gives admins a choice. IBeeRS are safe and cause headaches, call depth
tracking is considered to be s(t)ufficiently safe.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111149.029587352@infradead.org
Since ftrace has trampolines, don't use thunks for the __fentry__ site
but instead require that every function called from there includes
accounting. This very much includes all the direct-call functions.
Additionally, ftrace uses ROP tricks in two places:
- return_to_handler(), and
- ftrace_regs_caller() when pt_regs->orig_ax is set by a direct-call.
return_to_handler() already uses a retpoline to replace an
indirect-jump to defeat IBT, since this is a jump-type retpoline, make
sure there is no accounting done and ALTERNATIVE the RET into a ret.
ftrace_regs_caller() does much the same and gets the same treatment.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.927545073@infradead.org
ftrace_regs_caller() uses a PUSH;RET pattern to tail-call into a
direct-call function, this unbalances the RSB, fix that.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.823216933@infradead.org
Remove the weird jumps to RET and simply use RET.
This then promotes ftrace_stub() to a real function; which becomes
important for kcfi.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.719080593@infradead.org
Ensure that calls in BPF jitted programs are emitting call depth accounting
when enabled to keep the call/return balanced. The return thunk jump is
already injected due to the earlier retbleed mitigations.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.615413406@infradead.org
Callthunks addresses on the stack would confuse the ORC unwinder. Handle
them correctly and tell ORC to proceed further down the stack.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.511637628@infradead.org
Since the pre-symbol function padding is an integral part of the
symbol make kallsyms report it as part of the symbol by reporting it
as sym-x instead of prev_sym+y.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.409656012@infradead.org
When indirect calls are switched to direct calls then it has to be ensured
that the call target is not the function, but the call thunk when call
depth tracking is enabled. But static calls are available before call
thunks have been set up.
Ensure a second run through the static call patching code after call thunks
have been created. When call thunks are not enabled this has no side
effects.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.306100465@infradead.org
Add a debuigfs mechanism to validate the accounting, e.g. vs. call/ret
balance and to gather statistics about the stuffing to call ratio.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111148.204285506@infradead.org
Ensure that retpolines do the proper call accounting so that the return
accounting works correctly.
Specifically; retpolines are used to replace both 'jmp *%reg' and
'call *%reg', however these two cases do not have the same accounting
requirements. Therefore split things up and provide two different
retpoline arrays for SKL.
The 'jmp *%reg' case needs no accounting, the
__x86_indirect_jump_thunk_array[] covers this. The retpoline is
changed to not use the return thunk; it's a simple call;ret construct.
[ strictly speaking it should do:
andq $(~0x1f), PER_CPU_VAR(__x86_call_depth)
but we can argue this can be covered by the fuzz we already have
in the accounting depth (12) vs the RSB depth (16) ]
The 'call *%reg' case does need accounting, the
__x86_indirect_call_thunk_array[] covers this. Again, this retpoline
avoids the use of the return-thunk, in this case to avoid double
accounting.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.996634749@infradead.org
To address the Intel SKL RSB underflow issue in software it's required to
do call depth tracking.
Provide a return thunk for call depth tracking on Intel SKL CPUs.
The tracking does not use a counter. It uses uses arithmetic shift
right on call entry and logical shift left on return.
The depth tracking variable is initialized to 0x8000.... when the call
depth is zero. The arithmetic shift right sign extends the MSB and
saturates after the 12th call. The shift count is 5 so the tracking covers
12 nested calls. On return the variable is shifted left logically so it
becomes zero again.
CALL RET
0: 0x8000000000000000 0x0000000000000000
1: 0xfc00000000000000 0xf000000000000000
...
11: 0xfffffffffffffff8 0xfffffffffffffc00
12: 0xffffffffffffffff 0xffffffffffffffe0
After a return buffer fill the depth is credited 12 calls before the next
stuffing has to take place.
There is a inaccuracy for situations like this:
10 calls
5 returns
3 calls
4 returns
3 calls
....
The shift count might cause this to be off by one in either direction, but
there is still a cushion vs. the RSB depth. The algorithm does not claim to
be perfect, but it should obfuscate the problem enough to make exploitation
extremly difficult.
The theory behind this is:
RSB is a stack with depth 16 which is filled on every call. On the return
path speculation "pops" entries to speculate down the call chain. Once the
speculative RSB is empty it switches to other predictors, e.g. the Branch
History Buffer, which can be mistrained by user space and misguide the
speculation path to a gadget.
Call depth tracking is designed to break this speculation path by stuffing
speculation trap calls into the RSB which are never getting a corresponding
return executed. This stalls the prediction path until it gets resteered,
The assumption is that stuffing at the 12th return is sufficient to break
the speculation before it hits the underflow and the fallback to the other
predictors. Testing confirms that it works. Johannes, one of the retbleed
researchers. tried to attack this approach but failed.
There is obviously no scientific proof that this will withstand future
research progress, but all we can do right now is to speculate about it.
The SAR/SHL usage was suggested by Andi Kleen.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.890071690@infradead.org
In preparation for call depth tracking on Intel SKL CPUs, make it possible
to patch in a SKL specific return thunk.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.680469665@infradead.org
As for the builtins create call thunks and patch the call sites to call the
thunk on Intel SKL CPUs for retbleed mitigation.
Note, that module init functions are ignored for sake of simplicity because
loading modules is not something which is done in high frequent loops and
the attacker has not really a handle on when this happens in order to
launch a matching attack. The depth tracking will still work for calls into
the builtins and because the call is not accounted it will underflow faster
and overstuff, but that's mitigated by the saturating counter and the side
effect is only temporary.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.575673066@infradead.org
Mitigating the Intel SKL RSB underflow issue in software requires to
track the call depth. That is every CALL and every RET need to be
intercepted and additional code injected.
The existing retbleed mitigations already include means of redirecting
RET to __x86_return_thunk; this can be re-purposed and RET can be
redirected to another function doing RET accounting.
CALL accounting will use the function padding introduced in prior
patches. For each CALL instruction, the destination symbol's padding
is rewritten to do the accounting and the CALL instruction is adjusted
to call into the padding.
This ensures only affected CPUs pay the overhead of this accounting.
Unaffected CPUs will leave the padding unused and have their 'JMP
__x86_return_thunk' replaced with an actual 'RET' instruction.
Objtool has been modified to supply a .call_sites section that lists
all the 'CALL' instructions. Additionally the paravirt instruction
sites are iterated since they will have been patched from an indirect
call to direct calls (or direct instructions in which case it'll be
ignored).
Module handling and the actual thunk code for SKL will be added in
subsequent steps.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.470877038@infradead.org
For the upcoming call thunk patching it's less ifdeffery when the data
structure is unconditionally available. The code can then be trivially
fenced off with IS_ENABLED().
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.367853167@infradead.org
paranoid_entry(), error_entry() and xen_error_entry() have to be
exempted from call accounting by thunk patching because they are
before UNTRAIN_RET.
Expose them so they are available in the alternative code.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.265598113@infradead.org
The upcoming call thunk patching must hold text_mutex and needs access to
text_poke_copy(), which takes text_mutex.
Provide a _locked postfixed variant to expose the inner workings.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.159977224@infradead.org
Intel SKL CPUs fall back to other predictors when the RSB underflows. The
only microcode mitigation is IBRS which is insanely expensive. It comes
with performance drops of up to 30% depending on the workload.
A way less expensive, but nevertheless horrible mitigation is to track the
call depth in software and overeagerly fill the RSB when returns underflow
the software counter.
Provide a configuration symbol and a CPU misfeature bit.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111147.056176424@infradead.org
Now that all functions are 16 byte aligned, add 16 bytes of NOP
padding in front of each function. This prepares things for software
call stack tracking and kCFI/FineIBT.
This significantly increases kernel .text size, around 5.1% on a
x86_64-defconfig-ish build.
However, per the random access argument used for alignment, these 16
extra bytes are code that wouldn't be used. Performance measurements
back this up by showing no significant performance regressions.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111146.950884492@infradead.org
In preparation for mitigating the Intel SKL RSB underflow issue in
software, add a new configuration symbol which allows to build the
required call thunk infrastructure conditionally.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111146.849523555@infradead.org
No point in having a call there. Spare the call/ret overhead.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111146.539578813@infradead.org
Teach objtool about STT_NOTYPE -> STT_FUNC+0 sibling calls. Doing do
allows slightly simpler .S files.
There is a slight complication in that we specifically do not want to
allow sibling calls from symbol holes (previously covered by STT_WEAK
symbols) -- such things exist where a weak function has a .cold
subfunction for example.
Additionally, STT_NOTYPE tail-calls are allowed to happen with a
modified stack frame, they don't need to obey the normal rules after
all.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Currently insn->func contains a instruction -> symbol link for
STT_FUNC symbols. A NULL value is assumed to mean STT_NOTYPE.
However, there are also instructions not covered by any symbol at all.
This can happen due to __weak symbols for example.
Since the current scheme cannot differentiate between no symbol and
STT_NOTYPE symbol, change things around. Make insn->sym point to any
symbol type such that !insn->sym means no symbol and add a helper
insn_func() that check the sym->type to retain the old functionality.
This then prepares the way to add code that depends on the distinction
between STT_NOTYPE and no symbol at all.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
A semi common pattern is where code checks if a code address is
within a specific range. All text addresses require either ENDBR or
ANNOTATE_ENDBR, however the ANNOTATE_NOENDBR past the range is
unnatural.
Instead, suppress this warning when this is exactly at the end of a
symbol that itself starts with either ENDBR/ANNOTATE_ENDBR.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111146.434642471@infradead.org
The current find_{symbol,func}_containing() functions are broken in
the face of overlapping symbols, exactly the case that is needed for a
new ibt/endbr supression.
Import interval_tree_generic.h into the tools tree and convert the
symbol tree to an interval tree to support proper range stabs.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220915111146.330203761@infradead.org
