| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Currently ARM64 extracts which specific sanitizer has caused a trap via
encoded data in the trap instruction. Clang on x86 currently encodes the
same data in the UD1 instruction but x86 handle_bug() and
is_valid_bugaddr() currently only look at UD2.
Bring x86 to parity with ARM64, similar to commit 25b84002afb9 ("arm64:
Support Clang UBSAN trap codes for better reporting"). See the llvm
links for information about the code generation.
Enable the reporting of UBSAN sanitizer details on x86 compiled with clang
when CONFIG_UBSAN_TRAP=y by analysing UD1 and retrieving the type immediate
which is encoded by the compiler after the UD1.
[ tglx: Simplified it by moving the printk() into handle_bug() ]
Signed-off-by: Gatlin Newhouse <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Acked-by: Peter Zijlstra (Intel) <[email protected]>
Cc: Kees Cook <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
Link: https://github.com/llvm/llvm-project/commit/c5978f42ec8e9#diff-bb68d7cd885f41cfc35843998b0f9f534adb60b415f647109e597ce448e92d9f
Link: https://github.com/llvm/llvm-project/blob/main/llvm/lib/Target/X86/X86InstrSystem.td#L27
|
|
The default paranoid setting was updated in commit 0161028b7c8a
("perf/core: Change the default paranoia level to 2") so this comment is
no longer true.
Signed-off-by: James Clark <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
|
|
Some uncore PMON registers are located in the MMIO space of the Host
Bridge and DRAM Controller device, which is located at D0:F0 for
Tiger Lake and later client generation.
Use D0:F0 as a default device. So it doesn't need to keep adding the
complete Device ID list for each generation anymore.
Signed-off-by: Zhenyu Wang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Kan Liang <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
LNL uncore imc freerunning counters keep same as previous HW.
Signed-off-by: Zhenyu Wang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Kan Liang <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
The uncore subsystem for Lunar Lake is similar to the previous
Meteor Lake. The uncore PerfMon registers are located at both
MSR and MMIO space.
The ARB and iMC are kept. There is no difference from the Meteor Lake.
Move the global control initialization to the first box of the CBOX.
The sNCU is moved to the MMIO space.
The HBO is newly added and only be accessed from the MMIO space.
Signed-off-by: Kan Liang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Some uncore PMON registers are located in the MMIO space. For the client
machine, the MMIO space is usually located at D0:F0 but in a different
BAR. For example, some uncore PMON registers are located in the SAF BAR,
not the MCHBAR in the Lunar Lake.
The current __uncore_imc_init_box() hard code the BAR information.
Factor out the uncore_get_box_mmio_addr() which uses the BAR information
as a parameter.
The only change is the error output message. The hardcode name 'MCHBAR'
is replaced by the offset of a BAR.
Add a new macro, MMIO_UNCORE_COMMON_OPS(), since the MMIO ops functions
are usually the same among different generations.
Signed-off-by: Kan Liang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
>From the perspective of the uncore PMU, the Arrow Lake is the same as
the previous Meteor Lake. The only difference is the event list, which
will be supported in the perf tool later.
Signed-off-by: Kan Liang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
When ident_pud_init() uses only GB pages to create identity maps, large
ranges of addresses not actually requested can be included in the resulting
table; a 4K request will map a full GB. This can include a lot of extra
address space past that requested, including areas marked reserved by the
BIOS. That allows processor speculation into reserved regions, that on UV
systems can cause system halts.
Only use GB pages when map creation requests include the full GB page of
space. Fall back to using smaller 2M pages when only portions of a GB page
are included in the request.
No attempt is made to coalesce mapping requests. If a request requires a
map entry at the 2M (pmd) level, subsequent mapping requests within the
same 1G region will also be at the pmd level, even if adjacent or
overlapping such requests could have been combined to map a full GB page.
Existing usage starts with larger regions and then adds smaller regions, so
this should not have any great consequence.
Signed-off-by: Steve Wahl <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Tested-by: Pavin Joseph <[email protected]>
Tested-by: Sarah Brofeldt <[email protected]>
Tested-by: Eric Hagberg <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
A kexec kernel boot failure is sometimes observed on AMD CPUs due to an
unmapped EFI config table array. This can be seen when "nogbpages" is on
the kernel command line, and has been observed as a full BIOS reboot rather
than a successful kexec.
This was also the cause of reported regressions attributed to Commit
7143c5f4cf20 ("x86/mm/ident_map: Use gbpages only where full GB page should
be mapped.") which was subsequently reverted.
To avoid this page fault, explicitly include the EFI config table array in
the kexec identity map.
Further explanation:
The following 2 commits caused the EFI config table array to be
accessed when enabling sev at kernel startup.
commit ec1c66af3a30 ("x86/compressed/64: Detect/setup SEV/SME features
earlier during boot")
commit c01fce9cef84 ("x86/compressed: Add SEV-SNP feature
detection/setup")
This is in the code that examines whether SEV should be enabled or not, so
it can even affect systems that are not SEV capable.
This may result in a page fault if the EFI config table array's address is
unmapped. Since the page fault occurs before the new kernel establishes its
own identity map and page fault routines, it is unrecoverable and kexec
fails.
Most often, this problem is not seen because the EFI config table array
gets included in the map by the luck of being placed at a memory address
close enough to other memory areas that *are* included in the map created
by kexec.
Both the "nogbpages" command line option and the "use gpbages only where
full GB page should be mapped" change greatly reduce the chance of being
included in the map by luck, which is why the problem appears.
Signed-off-by: Tao Liu <[email protected]>
Signed-off-by: Steve Wahl <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Tested-by: Pavin Joseph <[email protected]>
Tested-by: Sarah Brofeldt <[email protected]>
Tested-by: Eric Hagberg <[email protected]>
Reviewed-by: Ard Biesheuvel <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Thomas Gleixner:
- Prevent a deadlock on cpu_hotplug_lock in the aperf/mperf driver.
A recent change in the ACPI code which consolidated code pathes moved
the invocation of init_freq_invariance_cppc() to be moved to a CPU
hotplug handler. The first invocation on AMD CPUs ends up enabling a
static branch which dead locks because the static branch enable tries
to acquire cpu_hotplug_lock but that lock is already held write by
the hotplug machinery.
Use static_branch_enable_cpuslocked() instead and take the hotplug
lock read for the Intel code path which is invoked from the
architecture code outside of the CPU hotplug operations.
- Fix the number of reserved bits in the sev_config structure bit field
so that the bitfield does not exceed 64 bit.
- Add missing Zen5 model numbers
- Fix the alignment assumptions of pti_clone_pgtable() and
clone_entry_text() on 32-bit:
The code assumes PMD aligned code sections, but on 32-bit the kernel
entry text is not PMD aligned. So depending on the code size and
location, which is configuration and compiler dependent, entry text
can cross a PMD boundary. As the start is not PMD aligned adding PMD
size to the start address is larger than the end address which
results in partially mapped entry code for user space. That causes
endless recursion on the first entry from userspace (usually #PF).
Cure this by aligning the start address in the addition so it ends up
at the next PMD start address.
clone_entry_text() enforces PMD mapping, but on 32-bit the tail might
eventually be PTE mapped, which causes a map fail because the PMD for
the tail is not a large page mapping. Use PTI_LEVEL_KERNEL_IMAGE for
the clone() invocation which resolves to PTE on 32-bit and PMD on
64-bit.
- Zero the 8-byte case for get_user() on range check failure on 32-bit
The recend consolidation of the 8-byte get_user() case broke the
zeroing in the failure case again. Establish it by clearing ECX
before the range check and not afterwards as that obvioulsy can't be
reached when the range check fails
* tag 'x86-urgent-2024-08-04' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/uaccess: Zero the 8-byte get_range case on failure on 32-bit
x86/mm: Fix pti_clone_entry_text() for i386
x86/mm: Fix pti_clone_pgtable() alignment assumption
x86/setup: Parse the builtin command line before merging
x86/CPU/AMD: Add models 0x60-0x6f to the Zen5 range
x86/sev: Fix __reserved field in sev_config
x86/aperfmperf: Fix deadlock on cpu_hotplug_lock
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 perf fixes from Thomas Gleixner:
- Move the smp_processor_id() invocation back into the non-preemtible
region, so that the result is valid to use
- Add the missing package C2 residency counters for Sierra Forest CPUs
to make the newly added support actually useful
* tag 'perf-urgent-2024-08-04' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/x86: Fix smp_processor_id()-in-preemptible warnings
perf/x86/intel/cstate: Add pkg C2 residency counter for Sierra Forest
|
|
A Linux guest on Hyper-V gets the TSC frequency from a synthetic MSR, if
available. In this case, set X86_FEATURE_TSC_KNOWN_FREQ so that Linux
doesn't unnecessarily do refined TSC calibration when setting up the TSC
clocksource.
With this change, a message such as this is no longer output during boot
when the TSC is used as the clocksource:
[ 1.115141] tsc: Refined TSC clocksource calibration: 2918.408 MHz
Furthermore, the guest and host will have exactly the same view of the
TSC frequency, which is important for features such as the TSC deadline
timer that are emulated by the Hyper-V host.
Signed-off-by: Michael Kelley <[email protected]>
Reviewed-by: Roman Kisel <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Wei Liu <[email protected]>
Message-ID: <[email protected]>
|
|
Pull kvm updates from Paolo Bonzini:
"The bulk of the changes here is a largish change to guest_memfd,
delaying the clearing and encryption of guest-private pages until they
are actually added to guest page tables. This started as "let's make
it impossible to misuse the API" for SEV-SNP; but then it ballooned a
bit.
The new logic is generally simpler and more ready for hugepage support
in guest_memfd.
Summary:
- fix latent bug in how usage of large pages is determined for
confidential VMs
- fix "underline too short" in docs
- eliminate log spam from limited APIC timer periods
- disallow pre-faulting of memory before SEV-SNP VMs are initialized
- delay clearing and encrypting private memory until it is added to
guest page tables
- this change also enables another small cleanup: the checks in
SNP_LAUNCH_UPDATE that limit it to non-populated, private pages can
now be moved in the common kvm_gmem_populate() function
- fix compilation error that the RISC-V merge introduced in selftests"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: x86/mmu: fix determination of max NPT mapping level for private pages
KVM: riscv: selftests: Fix compile error
KVM: guest_memfd: abstract how prepared folios are recorded
KVM: guest_memfd: let kvm_gmem_populate() operate only on private gfns
KVM: extend kvm_range_has_memory_attributes() to check subset of attributes
KVM: cleanup and add shortcuts to kvm_range_has_memory_attributes()
KVM: guest_memfd: move check for already-populated page to common code
KVM: remove kvm_arch_gmem_prepare_needed()
KVM: guest_memfd: make kvm_gmem_prepare_folio() operate on a single struct kvm
KVM: guest_memfd: delay kvm_gmem_prepare_folio() until the memory is passed to the guest
KVM: guest_memfd: return locked folio from __kvm_gmem_get_pfn
KVM: rename CONFIG_HAVE_KVM_GMEM_* to CONFIG_HAVE_KVM_ARCH_GMEM_*
KVM: guest_memfd: do not go through struct page
KVM: guest_memfd: delay folio_mark_uptodate() until after successful preparation
KVM: guest_memfd: return folio from __kvm_gmem_get_pfn()
KVM: x86: disallow pre-fault for SNP VMs before initialization
KVM: Documentation: Fix title underline too short warning
KVM: x86: Eliminate log spam from limited APIC timer periods
|
|
The unsynchronized_tsc() eventually checks num_possible_cpus(), and if the
system is non-Intel and the number of possible CPUs is greater than one,
assumes that TSCs are unsynchronized. This despite the comment saying
"assume multi socket systems are not synchronized", that is, socket rather
than CPU. This behavior was preserved by commit 8fbbc4b45ce3 ("x86: merge
tsc_init and clocksource code") and by the previous relevant commit
7e69f2b1ead2 ("clocksource: Remove the update callback").
The clocksource drivers were added by commit 5d0cf410e94b ("Time: i386
Clocksource Drivers") back in 2006, and the comment still said "socket"
rather than "CPU".
Therefore, bravely (and perhaps foolishly) make the code match the
comment.
Note that it is possible to bypass both code and comment by booting
with tsc=reliable, but this also disables the clocksource watchdog,
which is undesirable when trust in the TSC is strictly limited.
Reported-by: Zhengxu Chen <[email protected]>
Reported-by: Danielle Costantino <[email protected]>
Signed-off-by: Paul E. McKenney <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
* fix latent bug in how usage of large pages is determined for
confidential VMs
* fix "underline too short" in docs
* eliminate log spam from limited APIC timer periods
* disallow pre-faulting of memory before SEV-SNP VMs are initialized
* delay clearing and encrypting private memory until it is added to
guest page tables
* this change also enables another small cleanup: the checks in
SNP_LAUNCH_UPDATE that limit it to non-populated, private pages
can now be moved in the common kvm_gmem_populate() function
|
|
According to the data sheet, writing the MODE register should stop the
counter (and thus the interrupts). This appears to work on real hardware,
at least modern Intel and AMD systems. It should also work on Hyper-V.
However, on some buggy virtual machines the mode change doesn't have any
effect until the counter is subsequently loaded (or perhaps when the IRQ
next fires).
So, set MODE 0 and then load the counter, to ensure that those buggy VMs
do the right thing and the interrupts stop. And then write MODE 0 *again*
to stop the counter on compliant implementations too.
Apparently, Hyper-V keeps firing the IRQ *repeatedly* even in mode zero
when it should only happen once, but the second MODE write stops that too.
Userspace test program (mostly written by tglx):
=====
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <stdint.h>
#include <sys/io.h>
static __always_inline void __out##bwl(type value, uint16_t port) \
{ \
asm volatile("out" #bwl " %" #bw "0, %w1" \
: : "a"(value), "Nd"(port)); \
} \
\
static __always_inline type __in##bwl(uint16_t port) \
{ \
type value; \
asm volatile("in" #bwl " %w1, %" #bw "0" \
: "=a"(value) : "Nd"(port)); \
return value; \
}
BUILDIO(b, b, uint8_t)
#define inb __inb
#define outb __outb
#define PIT_MODE 0x43
#define PIT_CH0 0x40
#define PIT_CH2 0x42
static int is8254;
static void dump_pit(void)
{
if (is8254) {
// Latch and output counter and status
outb(0xC2, PIT_MODE);
printf("%02x %02x %02x\n", inb(PIT_CH0), inb(PIT_CH0), inb(PIT_CH0));
} else {
// Latch and output counter
outb(0x0, PIT_MODE);
printf("%02x %02x\n", inb(PIT_CH0), inb(PIT_CH0));
}
}
int main(int argc, char* argv[])
{
int nr_counts = 2;
if (argc > 1)
nr_counts = atoi(argv[1]);
if (argc > 2)
is8254 = 1;
if (ioperm(0x40, 4, 1) != 0)
return 1;
dump_pit();
printf("Set oneshot\n");
outb(0x38, PIT_MODE);
outb(0x00, PIT_CH0);
outb(0x0F, PIT_CH0);
dump_pit();
usleep(1000);
dump_pit();
printf("Set periodic\n");
outb(0x34, PIT_MODE);
outb(0x00, PIT_CH0);
outb(0x0F, PIT_CH0);
dump_pit();
usleep(1000);
dump_pit();
dump_pit();
usleep(100000);
dump_pit();
usleep(100000);
dump_pit();
printf("Set stop (%d counter writes)\n", nr_counts);
outb(0x30, PIT_MODE);
while (nr_counts--)
outb(0xFF, PIT_CH0);
dump_pit();
usleep(100000);
dump_pit();
usleep(100000);
dump_pit();
printf("Set MODE 0\n");
outb(0x30, PIT_MODE);
dump_pit();
usleep(100000);
dump_pit();
usleep(100000);
dump_pit();
return 0;
}
=====
Suggested-by: Sean Christopherson <[email protected]>
Co-developed-by: Li RongQing <[email protected]>
Signed-off-by: Li RongQing <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Tested-by: Michael Kelley <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
Leaving the PIT interrupt running can cause noticeable steal time for
virtual guests. The VMM generally has a timer which toggles the IRQ input
to the PIC and I/O APIC, which takes CPU time away from the guest. Even
on real hardware, running the counter may use power needlessly (albeit
not much).
Make sure it's turned off if it isn't going to be used.
Signed-off-by: David Woodhouse <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Tested-by: Michael Kelley <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
Despite multiple attempts to get the syscall number assignment right
for the newly added uretprobe syscall, we ended up with a bit of a mess:
- The number is defined as 467 based on the assumption that the
xattrat family of syscalls would use 463 through 466, but those
did not make it into 6.11.
- The include/uapi/asm-generic/unistd.h file still lists the number
463, but the new scripts/syscall.tbl that was supposed to have the
same data lists 467 instead as the number for arc, arm64, csky,
hexagon, loongarch, nios2, openrisc and riscv. None of these
architectures actually provide a uretprobe syscall.
- All the other architectures (powerpc, arm, mips, ...) don't list
this syscall at all.
There are two ways to make it consistent again: either list it with
the same syscall number on all architectures, or only list it on x86
but not in scripts/syscall.tbl and asm-generic/unistd.h.
Based on the most recent discussion, it seems like we won't need it
anywhere else, so just remove the inconsistent assignment and instead
move the x86 number to the next available one in the architecture
specific range, which is 335.
Fixes: 5c28424e9a34 ("syscalls: Fix to add sys_uretprobe to syscall.tbl")
Fixes: 190fec72df4a ("uprobe: Wire up uretprobe system call")
Fixes: 63ded110979b ("uprobe: Change uretprobe syscall scope and number")
Acked-by: Masami Hiramatsu (Google) <[email protected]>
Reviewed-by: Jiri Olsa <[email protected]>
Signed-off-by: Arnd Bergmann <[email protected]>
|
|
A process can disable access to the alternate signal stack by not
enabling the altstack's PKEY in the PKRU register.
Nevertheless, the kernel updates the PKRU temporarily for signal
handling. However, in sigreturn(), restore_sigcontext() will restore the
PKRU to the user-defined PKRU value.
This will cause restore_altstack() to fail with a SIGSEGV as it needs read
access to the altstack which is prohibited by the user-defined PKRU value.
Fix this by restoring altstack before restoring PKRU.
Signed-off-by: Aruna Ramakrishna <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
If the alternate signal stack is protected by a different PKEY than the
current execution stack, copying XSAVE data to the sigaltstack will fail
if its PKEY is not enabled in the PKRU register.
It's unknown which pkey was used by the application for the altstack, so
enable all PKEYS before XSAVE.
But this updated PKRU value is also pushed onto the sigframe, which
means the register value restored from sigcontext will be different from
the user-defined one, which is incorrect.
Fix that by overwriting the PKRU value on the sigframe with the original,
user-defined PKRU.
Signed-off-by: Aruna Ramakrishna <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
In the case where a user thread sets up an alternate signal stack protected
by the default PKEY (i.e. PKEY 0), while the thread's stack is protected by
a non-zero PKEY, both these PKEYS have to be enabled in the PKRU register
for the signal to be delivered to the application correctly. However, the
PKRU value restored after handling the signal must not enable this extra
PKEY (i.e. PKEY 0) - i.e., the PKRU value in the sigframe has to be
overwritten with the user-defined value.
Add helper functions that will update PKRU value in the sigframe after
XSAVE.
Note that sig_prepare_pkru() makes no assumption about which PKEY could
be used to protect the altstack (i.e. it may not be part of init_pkru),
and so enables all PKEYS.
No functional change.
Signed-off-by: Aruna Ramakrishna <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
Assume there's a multithreaded application that runs untrusted user
code. Each thread has its stack/code protected by a non-zero PKEY, and the
PKRU register is set up such that only that particular non-zero PKEY is
enabled. Each thread also sets up an alternate signal stack to handle
signals, which is protected by PKEY zero. The PKEYs man page documents that
the PKRU will be reset to init_pkru when the signal handler is invoked,
which means that PKEY zero access will be enabled. But this reset happens
after the kernel attempts to push fpu state to the alternate stack, which
is not (yet) accessible by the kernel, which leads to a new SIGSEGV being
sent to the application, terminating it.
Enabling both the non-zero PKEY (for the thread) and PKEY zero in
userspace will not work for this use case. It cannot have the alt stack
writeable by all - the rationale here is that the code running in that
thread (using a non-zero PKEY) is untrusted and should not have access
to the alternate signal stack (that uses PKEY zero), to prevent the
return address of a function from being changed. The expectation is that
kernel should be able to set up the alternate signal stack and deliver
the signal to the application even if PKEY zero is explicitly disabled
by the application. The signal handler accessibility should not be
dictated by whatever PKRU value the thread sets up.
The PKRU register is managed by XSAVE, which means the sigframe contents
must match the register contents - which is not the case here. It's
required that the signal frame contains the user-defined PKRU value (so
that it is restored correctly from sigcontext) but the actual register must
be reset to init_pkru so that the alt stack is accessible and the signal
can be delivered to the application. It seems that the proper fix here
would be to remove PKRU from the XSAVE framework and manage it separately,
which is quite complicated. As a workaround, do this:
orig_pkru = rdpkru();
wrpkru(orig_pkru & init_pkru_value);
xsave_to_user_sigframe();
put_user(pkru_sigframe_addr, orig_pkru)
In preparation for writing PKRU to sigframe, pass PKRU as an additional
parameter down the call chain from get_sigframe().
No functional change.
Signed-off-by: Aruna Ramakrishna <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
Bring x86 and selftests up to date
|
|
When tracing user functions with uprobe functionality, it's common to
install the probe (e.g., a BPF program) at the first instruction of the
function. This is often going to be `push %rbp` instruction in function
preamble, which means that within that function frame pointer hasn't
been established yet. This leads to consistently missing an actual
caller of the traced function, because perf_callchain_user() only
records current IP (capturing traced function) and then following frame
pointer chain (which would be caller's frame, containing the address of
caller's caller).
So when we have target_1 -> target_2 -> target_3 call chain and we are
tracing an entry to target_3, captured stack trace will report
target_1 -> target_3 call chain, which is wrong and confusing.
This patch proposes a x86-64-specific heuristic to detect `push %rbp`
(`push %ebp` on 32-bit architecture) instruction being traced. Given
entire kernel implementation of user space stack trace capturing works
under assumption that user space code was compiled with frame pointer
register (%rbp/%ebp) preservation, it seems pretty reasonable to use
this instruction as a strong indicator that this is the entry to the
function. In that case, return address is still pointed to by %rsp/%esp,
so we fetch it and add to stack trace before proceeding to unwind the
rest using frame pointer-based logic.
We also check for `endbr64` (for 64-bit modes) as another common pattern
for function entry, as suggested by Josh Poimboeuf. Even if we get this
wrong sometimes for uprobes attached not at the function entry, it's OK
because stack trace will still be overall meaningful, just with one
extra bogus entry. If we don't detect this, we end up with guaranteed to
be missing caller function entry in the stack trace, which is worse
overall.
Signed-off-by: Andrii Nakryiko <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
|
|
While zeroing the upper 32 bits of an 8-byte getuser on 32-bit x86 was
fixed by commit 8c860ed825cb ("x86/uaccess: Fix missed zeroing of ia32 u64
get_user() range checking") it was broken again in commit 8a2462df1547
("x86/uaccess: Improve the 8-byte getuser() case").
This is because the register which holds the upper 32 bits (%ecx) is being
cleared _after_ the check_range, so if the range check fails, %ecx is never
cleared.
This can be reproduced with:
./tools/testing/kunit/kunit.py run --arch i386 usercopy
Instead, clear %ecx _before_ check_range in the 8-byte case. This
reintroduces a bit of the ugliness we were trying to avoid by adding
another #ifndef CONFIG_X86_64, but at least keeps check_range from needing
a separate bad_get_user_8 jump.
Fixes: 8a2462df1547 ("x86/uaccess: Improve the 8-byte getuser() case")
Signed-off-by: David Gow <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Acked-by: Linus Torvalds <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
|
|
The `if (req_max_level)` test was meant ignore req_max_level if
PG_LEVEL_NONE was returned. Hence, this function should return
max_level instead of the ignored req_max_level.
This is only a latent issue for now, since guest_memfd does not
support large pages.
Signed-off-by: Ackerley Tng <[email protected]>
Message-ID: <[email protected]>
Fixes: f32fb32820b1 ("KVM: x86: Add hook for determining max NPT mapping level")
Signed-off-by: Paolo Bonzini <[email protected]>
|
|
Current AMD systems can report MCA errors using the ACPI Boot Error
Record Table (BERT). The BERT entries for MCA errors will be an x86
Common Platform Error Record (CPER) with an MSR register context that
matches the MCAX/SMCA register space.
However, the BERT will not necessarily be processed on the CPU that
reported the MCA errors. Therefore, the correct CPU number needs to be
determined and the information saved in struct mce.
Use the newly defined mce_prep_record_*() helpers to get the correct
data.
Also, add an explicit check to verify that a valid CPU number was found
from the APIC ID search.
Signed-off-by: Yazen Ghannam <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Reviewed-by: Nikolay Borisov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
|
|
Generally, MCA information for an error is gathered on the CPU that
reported the error. In this case, CPU-specific information from the
running CPU will be correct.
However, this will be incorrect if the MCA information is gathered while
running on a CPU that didn't report the error. One example is creating
an MCA record using mce_prep_record() for errors reported from ACPI.
Split mce_prep_record() so that there is a helper function to gather
common, i.e. not CPU-specific, information and another helper for
CPU-specific information.
Leave mce_prep_record() defined as-is for the common case when running
on the reporting CPU.
Get MCG_CAP in the global helper even though the register is per-CPU.
This value is not already cached per-CPU like other values. And it does
not assist with any per-CPU decoding or handling.
Signed-off-by: Yazen Ghannam <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Reviewed-by: Nikolay Borisov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
|
|
There is no MCE "setup" done in mce_setup(). Rather, this function initializes
and prepares an MCE record.
Rename the function to highlight what it does.
No functional change is intended.
Suggested-by: Borislav Petkov <[email protected]>
Signed-off-by: Yazen Ghannam <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Reviewed-by: Nikolay Borisov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
|
|
While x86_64 has PMD aligned text sections, i386 does not have this
luxery. Notably ALIGN_ENTRY_TEXT_END is empty and _etext has PAGE
alignment.
This means that text on i386 can be page granular at the tail end,
which in turn means that the PTI text clones should consistently
account for this.
Make pti_clone_entry_text() consistent with pti_clone_kernel_text().
Fixes: 16a3fe634f6a ("x86/mm/pti: Clone kernel-image on PTE level for 32 bit")
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
|
|
Guenter reported dodgy crashes on an i386-nosmp build using GCC-11
that had the form of endless traps until entry stack exhaust and then
#DF from the stack guard.
It turned out that pti_clone_pgtable() had alignment assumptions on
the start address, notably it hard assumes start is PMD aligned. This
is true on x86_64, but very much not true on i386.
These assumptions can cause the end condition to malfunction, leading
to a 'short' clone. Guess what happens when the user mapping has a
short copy of the entry text?
Use the correct increment form for addr to avoid alignment
assumptions.
Fixes: 16a3fe634f6a ("x86/mm/pti: Clone kernel-image on PTE level for 32 bit")
Reported-by: Guenter Roeck <[email protected]>
Tested-by: Guenter Roeck <[email protected]>
Suggested-by: Thomas Gleixner <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
|
|
Commit in Fixes was added as a catch-all for cases where the cmdline is
parsed before being merged with the builtin one.
And promptly one issue appeared, see Link below. The microcode loader
really needs to parse it that early, but the merging happens later.
Reshuffling the early boot nightmare^W code to handle that properly would
be a painful exercise for another day so do the chicken thing and parse the
builtin cmdline too before it has been merged.
Fixes: 0c40b1c7a897 ("x86/setup: Warn when option parsing is done too early")
Reported-by: Mike Lothian <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/20240730152108.GAZqkE5Dfi9AuKllRw@fat_crate.local
Link: https://lore.kernel.org/r/20240722152330.GCZp55ck8E_FT4kPnC@fat_crate.local
|
|
Commit b50db7095fe0 ("x86/tsc: Disable clocksource watchdog for TSC on
qualified platorms") was introduced to solve problem that sometimes TSC
clocksource is wrongly judged as unstable by watchdog like 'jiffies', HPET,
etc.
In it, the hardware package number is a key factor for judging whether to
disable the watchdog for TSC, and 'nr_online_nodes' was chosen due to, at
that time (kernel v5.1x), it is available in early boot phase before
registering 'tsc-early' clocksource, where all non-boot CPUs are not
brought up yet.
Dave and Rui pointed out there are many cases in which 'nr_online_nodes'
is cheated and not accurate, like:
* SNC (sub-numa cluster) mode enabled
* numa emulation (numa=fake=8 etc.)
* numa=off
* platforms with CPU-less HBM nodes, CPU-less Optane memory nodes.
* 'maxcpus=' cmdline setup, where chopped CPUs could be onlined later
* 'nr_cpus=', 'possible_cpus=' cmdline setup, where chopped CPUs can
not be onlined after boot
The SNC case is the most user-visible case, as many CSP (Cloud Service
Provider) enable this feature in their server fleets. When SNC3 enabled, a
2 socket machine will appear to have 6 NUMA nodes, and get impacted by the
issue in reality.
Thomas' recent patchset of refactoring x86 topology code improves
topology_max_packages() greatly, by making it more accurate and available
in early boot phase, which works well in most of the above cases.
The only exceptions are 'nr_cpus=' and 'possible_cpus=' setup, which may
under-estimate the package number. As during topology setup, the boot CPU
iterates through all enumerated APIC IDs and either accepts or rejects the
APIC ID. For accepted IDs, it figures out which bits of the ID map to the
package number. It tracks which package numbers have been seen in a
bitmap. topology_max_packages() just returns the number of bits set in
that bitmap.
'nr_cpus=' and 'possible_cpus=' can cause more APIC IDs to be rejected and
can artificially lower the number of bits in the package bitmap and thus
topology_max_packages(). This means that, for example, a system with 8
physical packages might reject all the CPUs on 6 of those packages and be
left with only 2 packages and 2 bits set in the package bitmap. It needs
the TSC watchdog, but would disable it anyway. This isn't ideal, but it
only happens for debug-oriented options. This is fixable by tracking the
package numbers for rejected CPUs. But it's not worth the trouble for
debugging.
So use topology_max_packages() to replace nr_online_nodes().
Reported-by: Dave Hansen <[email protected]>
Signed-off-by: Feng Tang <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Reviewed-by: Waiman Long <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
Closes: https://lore.kernel.org/lkml/[email protected]/
|
|
The following bug was triggered on a system built with
CONFIG_DEBUG_PREEMPT=y:
# echo p > /proc/sysrq-trigger
BUG: using smp_processor_id() in preemptible [00000000] code: sh/117
caller is perf_event_print_debug+0x1a/0x4c0
CPU: 3 UID: 0 PID: 117 Comm: sh Not tainted 6.11.0-rc1 #109
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1.1 04/01/2014
Call Trace:
<TASK>
dump_stack_lvl+0x4f/0x60
check_preemption_disabled+0xc8/0xd0
perf_event_print_debug+0x1a/0x4c0
__handle_sysrq+0x140/0x180
write_sysrq_trigger+0x61/0x70
proc_reg_write+0x4e/0x70
vfs_write+0xd0/0x430
? handle_mm_fault+0xc8/0x240
ksys_write+0x9c/0xd0
do_syscall_64+0x96/0x190
entry_SYSCALL_64_after_hwframe+0x4b/0x53
This is because the commit d4b294bf84db ("perf/x86: Hybrid PMU support
for counters") took smp_processor_id() outside the irq critical section.
If a preemption occurs in perf_event_print_debug() and the task is
migrated to another cpu, we may get incorrect pmu debug information.
Move smp_processor_id() back inside the irq critical section to fix this
issue.
Fixes: d4b294bf84db ("perf/x86: Hybrid PMU support for counters")
Signed-off-by: Li Huafei <[email protected]>
Reviewed-and-tested-by: K Prateek Nayak <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Kan Liang <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Add some new Zen5 models for the 0x1A family.
[ bp: Merge the 0x60 and 0x70 ranges. ]
Signed-off-by: Perry Yuan <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated, where some
mitigations have entries in Kconfig, and they could be modified, while others
mitigations do not have Kconfig entries, and could not be controlled at build
time.
Create a new kernel config that allows GDS to be completely disabled,
similarly to the "gather_data_sampling=off" or "mitigations=off" kernel
command-line.
Now, there are two options for GDS mitigation:
* CONFIG_MITIGATION_GDS=n -> Mitigation disabled (New)
* CONFIG_MITIGATION_GDS=y -> Mitigation enabled (GDS_MITIGATION_FULL)
Suggested-by: Josh Poimboeuf <[email protected]>
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Remove the MITIGATION_GDS_FORCE Kconfig option, which aggressively disables
AVX as a mitigation for Gather Data Sampling (GDS) vulnerabilities. This
option is not widely used by distros.
While removing the Kconfig option, retain the runtime configuration ability
through the `gather_data_sampling=force` kernel parameter. This allows users
to still enable this aggressive mitigation if needed, without baking it into
the kernel configuration.
Simplify the kernel configuration while maintaining flexibility for runtime
mitigation choices.
Suggested-by: Borislav Petkov <[email protected]>
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Reviewed-by: Daniel Sneddon <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the SSB CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the Spectre V2 CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the SRBDS CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the Spectre v1 CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the RETBLEED CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the L1TF CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the MMIO Stale data CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the TAA CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Currently, the CONFIG_SPECULATION_MITIGATIONS is halfway populated,
where some mitigations have entries in Kconfig, and they could be
modified, while others mitigations do not have Kconfig entries, and
could not be controlled at build time.
Create an entry for the MDS CPU mitigation under
CONFIG_SPECULATION_MITIGATIONS. This allow users to enable or disable
it at compilation time.
Signed-off-by: Breno Leitao <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Acked-by: Josh Poimboeuf <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
sev_config currently has debug, ghcbs_initialized, and use_cas fields.
However, __reserved count has not been updated. Fix this.
Fixes: 34ff65901735 ("x86/sev: Use kernel provided SVSM Calling Areas")
Signed-off-by: Pavan Kumar Paluri <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|
|
Initialize equiv_id in order to shut up:
arch/x86/kernel/cpu/microcode/amd.c:714:6: warning: variable 'equiv_id' is \
used uninitialized whenever 'if' condition is false [-Wsometimes-uninitialized]
if (x86_family(bsp_cpuid_1_eax) < 0x17) {
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
because clang doesn't do interprocedural analysis for warnings to see
that this variable won't be used uninitialized.
Fixes: 94838d230a6c ("x86/microcode/AMD: Use the family,model,stepping encoded in the patch ID")
Reported-by: kernel test robot <[email protected]>
Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
|
|
This patch fixes a tailcall issue caused by abusing the tailcall in
bpf2bpf feature.
As we know, tail_call_cnt propagates by rax from caller to callee when
to call subprog in tailcall context. But, like the following example,
MAX_TAIL_CALL_CNT won't work because of missing tail_call_cnt
back-propagation from callee to caller.
\#include <linux/bpf.h>
\#include <bpf/bpf_helpers.h>
\#include "bpf_legacy.h"
struct {
__uint(type, BPF_MAP_TYPE_PROG_ARRAY);
__uint(max_entries, 1);
__uint(key_size, sizeof(__u32));
__uint(value_size, sizeof(__u32));
} jmp_table SEC(".maps");
int count = 0;
static __noinline
int subprog_tail1(struct __sk_buff *skb)
{
bpf_tail_call_static(skb, &jmp_table, 0);
return 0;
}
static __noinline
int subprog_tail2(struct __sk_buff *skb)
{
bpf_tail_call_static(skb, &jmp_table, 0);
return 0;
}
SEC("tc")
int entry(struct __sk_buff *skb)
{
volatile int ret = 1;
count++;
subprog_tail1(skb);
subprog_tail2(skb);
return ret;
}
char __license[] SEC("license") = "GPL";
At run time, the tail_call_cnt in entry() will be propagated to
subprog_tail1() and subprog_tail2(). But, when the tail_call_cnt in
subprog_tail1() updates when bpf_tail_call_static(), the tail_call_cnt
in entry() won't be updated at the same time. As a result, in entry(),
when tail_call_cnt in entry() is less than MAX_TAIL_CALL_CNT and
subprog_tail1() returns because of MAX_TAIL_CALL_CNT limit,
bpf_tail_call_static() in suprog_tail2() is able to run because the
tail_call_cnt in subprog_tail2() propagated from entry() is less than
MAX_TAIL_CALL_CNT.
So, how many tailcalls are there for this case if no error happens?
From top-down view, does it look like hierarchy layer and layer?
With this view, there will be 2+4+8+...+2^33 = 2^34 - 2 = 17,179,869,182
tailcalls for this case.
How about there are N subprog_tail() in entry()? There will be almost
N^34 tailcalls.
Then, in this patch, it resolves this case on x86_64.
In stead of propagating tail_call_cnt from caller to callee, it
propagates its pointer, tail_call_cnt_ptr, tcc_ptr for short.
However, where does it store tail_call_cnt?
It stores tail_call_cnt on the stack of main prog. When tail call
happens in subprog, it increments tail_call_cnt by tcc_ptr.
Meanwhile, it stores tail_call_cnt_ptr on the stack of main prog, too.
And, before jump to tail callee, it has to pop tail_call_cnt and
tail_call_cnt_ptr.
Then, at the prologue of subprog, it must not make rax as
tail_call_cnt_ptr again. It has to reuse tail_call_cnt_ptr from caller.
As a result, at run time, it has to recognize rax is tail_call_cnt or
tail_call_cnt_ptr at prologue by:
1. rax is tail_call_cnt if rax is <= MAX_TAIL_CALL_CNT.
2. rax is tail_call_cnt_ptr if rax is > MAX_TAIL_CALL_CNT, because a
pointer won't be <= MAX_TAIL_CALL_CNT.
Here's an example to dump JITed.
struct {
__uint(type, BPF_MAP_TYPE_PROG_ARRAY);
__uint(max_entries, 1);
__uint(key_size, sizeof(__u32));
__uint(value_size, sizeof(__u32));
} jmp_table SEC(".maps");
int count = 0;
static __noinline
int subprog_tail(struct __sk_buff *skb)
{
bpf_tail_call_static(skb, &jmp_table, 0);
return 0;
}
SEC("tc")
int entry(struct __sk_buff *skb)
{
int ret = 1;
count++;
subprog_tail(skb);
subprog_tail(skb);
return ret;
}
When bpftool p d j id 42:
int entry(struct __sk_buff * skb):
bpf_prog_0c0f4c2413ef19b1_entry:
; int entry(struct __sk_buff *skb)
0: endbr64
4: nopl (%rax,%rax)
9: xorq %rax, %rax ;; rax = 0 (tail_call_cnt)
c: pushq %rbp
d: movq %rsp, %rbp
10: endbr64
14: cmpq $33, %rax ;; if rax > 33, rax = tcc_ptr
18: ja 0x20 ;; if rax > 33 goto 0x20 ---+
1a: pushq %rax ;; [rbp - 8] = rax = 0 |
1b: movq %rsp, %rax ;; rax = rbp - 8 |
1e: jmp 0x21 ;; ---------+ |
20: pushq %rax ;; <--------|---------------+
21: pushq %rax ;; <--------+ [rbp - 16] = rax
22: pushq %rbx ;; callee saved
23: movq %rdi, %rbx ;; rbx = skb (callee saved)
; count++;
26: movabsq $-82417199407104, %rdi
30: movl (%rdi), %esi
33: addl $1, %esi
36: movl %esi, (%rdi)
; subprog_tail(skb);
39: movq %rbx, %rdi ;; rdi = skb
3c: movq -16(%rbp), %rax ;; rax = tcc_ptr
43: callq 0x80 ;; call subprog_tail()
; subprog_tail(skb);
48: movq %rbx, %rdi ;; rdi = skb
4b: movq -16(%rbp), %rax ;; rax = tcc_ptr
52: callq 0x80 ;; call subprog_tail()
; return ret;
57: movl $1, %eax
5c: popq %rbx
5d: leave
5e: retq
int subprog_tail(struct __sk_buff * skb):
bpf_prog_3a140cef239a4b4f_subprog_tail:
; int subprog_tail(struct __sk_buff *skb)
0: endbr64
4: nopl (%rax,%rax)
9: nopl (%rax) ;; do not touch tail_call_cnt
c: pushq %rbp
d: movq %rsp, %rbp
10: endbr64
14: pushq %rax ;; [rbp - 8] = rax (tcc_ptr)
15: pushq %rax ;; [rbp - 16] = rax (tcc_ptr)
16: pushq %rbx ;; callee saved
17: pushq %r13 ;; callee saved
19: movq %rdi, %rbx ;; rbx = skb
; asm volatile("r1 = %[ctx]\n\t"
1c: movabsq $-105487587488768, %r13 ;; r13 = jmp_table
26: movq %rbx, %rdi ;; 1st arg, skb
29: movq %r13, %rsi ;; 2nd arg, jmp_table
2c: xorl %edx, %edx ;; 3rd arg, index = 0
2e: movq -16(%rbp), %rax ;; rax = [rbp - 16] (tcc_ptr)
35: cmpq $33, (%rax)
39: jae 0x4e ;; if *tcc_ptr >= 33 goto 0x4e --------+
3b: jmp 0x4e ;; jmp bypass, toggled by poking |
40: addq $1, (%rax) ;; (*tcc_ptr)++ |
44: popq %r13 ;; callee saved |
46: popq %rbx ;; callee saved |
47: popq %rax ;; undo rbp-16 push |
48: popq %rax ;; undo rbp-8 push |
49: nopl (%rax,%rax) ;; tail call target, toggled by poking |
; return 0; ;; |
4e: popq %r13 ;; restore callee saved <--------------+
50: popq %rbx ;; restore callee saved
51: leave
52: retq
Furthermore, when trampoline is the caller of bpf prog, which is
tail_call_reachable, it is required to propagate rax through trampoline.
Fixes: ebf7d1f508a7 ("bpf, x64: rework pro/epilogue and tailcall handling in JIT")
Fixes: e411901c0b77 ("bpf: allow for tailcalls in BPF subprograms for x64 JIT")
Reviewed-by: Eduard Zingerman <[email protected]>
Signed-off-by: Leon Hwang <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Alexei Starovoitov <[email protected]>
Signed-off-by: Andrii Nakryiko <[email protected]>
|
|
The broken patch results in a call to init_freq_invariance_cppc() in a CPU
hotplug handler in both the path for initially present CPUs and those
hotplugged later. That function includes a one time call to
amd_set_max_freq_ratio() which in turn calls freq_invariance_enable() that has
a static_branch_enable() which takes the cpu_hotlug_lock which is already
held.
Avoid the deadlock by using static_branch_enable_cpuslocked() as the lock will
always be already held. The equivalent path on Intel does not already hold
this lock, so take it around the call to freq_invariance_enable(), which
results in it being held over the call to register_syscall_ops, which looks to
be safe to do.
Fixes: c1385c1f0ba3 ("ACPI: processor: Simplify initial onlining to use same path for cold and hotplug")
Closes: https://lore.kernel.org/all/CABXGCsPvqBfL5hQDOARwfqasLRJ_eNPBbCngZ257HOe=xbWDkA@mail.gmail.com/
Reported-by: Mikhail Gavrilov <[email protected]>
Suggested-by: Thomas Gleixner <[email protected]>
Signed-off-by: Jonathan Cameron <[email protected]>
Signed-off-by: Borislav Petkov (AMD) <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Tested-by: Mikhail Gavrilov <[email protected]>
Tested-by: Borislav Petkov (AMD) <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
|