diff options
author | 2024-09-18 16:01:22 +0200 | |
---|---|---|
committer | 2024-09-27 20:25:36 +0200 | |
commit | f766fb0a85fa06a9081e6b39c0834960202fde85 (patch) | |
tree | be4f80a3dbe43ffc2b6447260722a89565a274f9 /sysdeps/unix/sysv/linux/getrandom.c | |
parent | elf: Fix tst-dlopen-tlsreinit1.out test dependency (diff) | |
download | glibc-vdso.tar.xz glibc-vdso.zip |
linux: Add support for getrandom vDSOvdso
Linux 6.11 has getrandom() in vDSO. It operates on a thread-local opaque
state allocated with mmap using flags specified by the vDSO.
Multiple states are allocated at once, as many as fit into a page, and
these are held in an array of available states to be doled out to each
thread upon first use, and recycled when a thread terminates. As these
states run low, more are allocated.
To make this procedure async-signal-safe, a simple guard is used in the
LSB of the opaque state address, falling back to the syscall if there's
reentrancy contention.
Also, _Fork() is handled by blocking signals on opaque state allocation
(so _Fork() always sees a consistent state even if it interrupts a
getrandom() call) and by iterating over the thread stack cache on
reclaim_stack. Each opaque state will be in the free states list
(grnd_alloc.states) or allocated to a running thread.
The cancellation is handled by always using GRND_NONBLOCK flags while
calling the vDSO, and falling back to the cancellable syscall if the
kernel returns EAGAIN (would block). Since getrandom is not defined by
POSIX and cancellation is supported as an extension, the cancellation is
handled as 'may occur' instead of 'shall occur' [1], meaning that if
vDSO does not block (the expected behavior) getrandom will not act as a
cancellation entrypoint. It avoids a pthread_testcancel call on the fast
path (different than 'shall occur' functions, like sem_wait()).
It is currently enabled for x86_64, which is available in Linux 6.11,
and aarch64, powerpc32, powerpc64, loongarch64, and s390x, which are
available in Linux 6.12.
Link: https://pubs.opengroup.org/onlinepubs/9799919799/nframe.html [1]
Co-developed-by: Jason A. Donenfeld <Jason@zx2c4.com>
Tested-by: Jason A. Donenfeld <Jason@zx2c4.com> # x86_64
Tested-by: Adhemerval Zanella <adhemerval.zanella@linaro.org> # x86_64, aarch64
Tested-by: Xi Ruoyao <xry111@xry111.site> # x86_64, aarch64, loongarch64
Tested-by: Stefan Liebler <stli@linux.ibm.com> # s390x
Diffstat (limited to 'sysdeps/unix/sysv/linux/getrandom.c')
-rw-r--r-- | sysdeps/unix/sysv/linux/getrandom.c | 299 |
1 files changed, 298 insertions, 1 deletions
diff --git a/sysdeps/unix/sysv/linux/getrandom.c b/sysdeps/unix/sysv/linux/getrandom.c index 777d1decf0..8030438f76 100644 --- a/sysdeps/unix/sysv/linux/getrandom.c +++ b/sysdeps/unix/sysv/linux/getrandom.c @@ -21,12 +21,309 @@ #include <unistd.h> #include <sysdep-cancel.h> +static inline ssize_t +getrandom_syscall (void *buffer, size_t length, unsigned int flags, + bool cancel) +{ + return cancel + ? SYSCALL_CANCEL (getrandom, buffer, length, flags) + : INLINE_SYSCALL_CALL (getrandom, buffer, length, flags); +} + +#ifdef HAVE_GETRANDOM_VSYSCALL +# include <assert.h> +# include <ldsodefs.h> +# include <libc-lock.h> +# include <list.h> +# include <setvmaname.h> +# include <sys/mman.h> +# include <sys/sysinfo.h> +# include <tls-internal.h> + +/* These values will be initialized at loading time by calling the] + _dl_vdso_getrandom with a special value. The 'state_size' is the opaque + state size per-thread allocated with a mmap using 'mmap_prot' and + 'mmap_flags' argument. */ +static uint32_t state_size; +static uint32_t state_size_cache_aligned; +static uint32_t mmap_prot; +static uint32_t mmap_flags; + +void +__getrandom_early_init (_Bool initial) +{ + if (initial && (GLRO (dl_vdso_getrandom) != NULL)) + { + /* Used to query the vDSO for the required mmap flags and the opaque + per-thread state size. Defined by linux/random.h. */ + struct vgetrandom_opaque_params + { + uint32_t size_of_opaque_state; + uint32_t mmap_prot; + uint32_t mmap_flags; + uint32_t reserved[13]; + } params; + if (GLRO(dl_vdso_getrandom) (NULL, 0, 0, ¶ms, ~0UL) == 0) + { + /* Align each opaque state to L1 data cache size to avoid false + sharing. If the size can not be obtained, use the kernel + provided one. */ + state_size = params.size_of_opaque_state; + + long int ld1sz = __sysconf (_SC_LEVEL1_DCACHE_LINESIZE) ?: 1; + state_size_cache_aligned = ALIGN_UP (state_size, ld1sz); + + mmap_prot = params.mmap_prot; + mmap_flags = params.mmap_flags; + } + } +} + +/* The function below are used on reentracy handling with (i.e. SA_NODEFER). + Befor allocate a new state or issue the vDSO, atomically read the current + thread buffer, and if this is already reserved (is_reserved_ptr) fallback + to the syscall. Otherwise, reserve the buffer by atomically setting the + LSB of the opaque state pointer. The bit is cleared after the vDSO is + called, or before issuing the fallback syscall. */ + +static inline void *reserve_ptr (void *p) +{ + return (void *) ((uintptr_t) (p) | 1UL); +} + +static inline void *release_ptr (void *p) +{ + return (void *) ((uintptr_t) (p) & ~1UL); +} + +static inline bool is_reserved_ptr (void *p) +{ + return (uintptr_t) (p) & 1UL; +} + +static struct +{ + __libc_lock_define (, lock); + + void **states; /* Queue of opaque states allocated with the kernel + provided flags and used on getrandom vDSO call. */ + size_t len; /* Number of available free states in the queue. */ + size_t total; /* Number of states allocated from the kernel. */ + size_t cap; /* Total numver of states that 'states' can hold before + needed to be resized. */ +} grnd_alloc = { + .lock = LLL_LOCK_INITIALIZER +}; + +static bool +vgetrandom_get_state_alloc (void) +{ + /* Start by allocating one page for the opaque states. */ + size_t block_size = ALIGN_UP (state_size_cache_aligned, GLRO(dl_pagesize)); + size_t num = GLRO (dl_pagesize) / state_size_cache_aligned; + void *block = __mmap (NULL, GLRO(dl_pagesize), mmap_prot, mmap_flags, -1, 0); + if (block == MAP_FAILED) + return false; + __set_vma_name (block, block_size, " glibc: getrandom"); + + if (grnd_alloc.total + num > grnd_alloc.cap) + { + /* Use a new mmap instead of trying to mremap. It avoids a + potential multithread fork issue where fork is called just after + mremap returns but before assigning to the grnd_alloc.states, + thus making the its value invalid in the child. */ + void *old_states = grnd_alloc.states; + size_t old_states_size = ALIGN_UP (sizeof (*grnd_alloc.states) * + grnd_alloc.total + num, + GLRO(dl_pagesize)); + size_t states_size; + if (old_states == NULL) + states_size = old_states_size; + else + states_size = ALIGN_UP (sizeof (*grnd_alloc.states) * grnd_alloc.cap, + GLRO(dl_pagesize)); + + /* There is no need to memcpy any opaque state information because + all the allocated opaque states are assigned to running threads + (meaning that if we iterate over them we can reconstruct the state + list). */ + void **states = __mmap (NULL, states_size, PROT_READ | PROT_WRITE, + MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); + if (states == MAP_FAILED) + { + __munmap (block, block_size); + return false; + } + + /* Atomically replace the old state, so if a fork happens the child + process will see a consistent free state buffer. The size might + not be updated, but it does not really matter since the buffer is + always increased. */ + grnd_alloc.states = states; + atomic_thread_fence_seq_cst (); + if (old_states != NULL) + __munmap (old_states, old_states_size); + + __set_vma_name (states, states_size, " glibc: getrandom states"); + grnd_alloc.cap = states_size / sizeof (*grnd_alloc.states); + atomic_thread_fence_seq_cst (); + } + + for (size_t i = 0; i < num; ++i) + { + /* States should not straddle a page, no need to use the aligned + value because this is just an optimization. */ + if (((uintptr_t) block & (GLRO (dl_pagesize) - 1)) + state_size + > GLRO (dl_pagesize)) + block = PTR_ALIGN_UP (block, GLRO(dl_pagesize)); + grnd_alloc.states[i] = block; + block += state_size_cache_aligned; + } + /* Concurrent fork should not observe the previous pointer value. */ + grnd_alloc.len = num; + grnd_alloc.total += num; + atomic_thread_fence_seq_cst (); + + return true; +} + +/* Allocate an opaque state for vgetrandom. If the grnd_alloc does not have + any, mmap() another page of them using the vgetrandom parameters. */ +static void * +vgetrandom_get_state (void) +{ + void *state = NULL; + + /* The signal blocking avoid the potential issue where _Fork() (which is + async-signal-safe) is called with the lock taken. The function is + called only once during thread lifetime, so the overhead should be + minimal. */ + internal_sigset_t set; + internal_signal_block_all (&set); + __libc_lock_lock (grnd_alloc.lock); + + if (grnd_alloc.len > 0 || vgetrandom_get_state_alloc ()) + state = grnd_alloc.states[--grnd_alloc.len]; + + __libc_lock_unlock (grnd_alloc.lock); + internal_signal_restore_set (&set); + + return state; +} + +/* Returns true when vgetrandom is used successfully. Returns false if the + syscall fallback should be issued in the case the vDSO is not present, in + the case of reentrancy, or if any memory allocation fails. */ +static ssize_t +getrandom_vdso (void *buffer, size_t length, unsigned int flags, bool cancel) +{ + if (__glibc_unlikely (state_size == 0)) + return getrandom_syscall (buffer, length, flags, cancel); + + struct pthread *self = THREAD_SELF; + + void *state = atomic_load_relaxed (&self->getrandom_buf); + if (is_reserved_ptr (state)) + return getrandom_syscall (buffer, length, flags, cancel); + atomic_store_relaxed (&self->getrandom_buf, reserve_ptr (state)); + + bool r = false; + if (state == NULL) + { + state = vgetrandom_get_state (); + if (state == NULL) + goto out; + } + + /* Since the vDSO implementation does not issue the syscall with the + cancellation bridge (__syscall_cancel_arch), use GRND_NONBLOCK so there + is no potential unbounded blocking in the kernel. It should be a rare + situation, only at system startup when RNG is not initialized. */ + ssize_t ret = GLRO (dl_vdso_getrandom) (buffer, + length, + flags | GRND_NONBLOCK, + state, + state_size); + if (INTERNAL_SYSCALL_ERROR_P (ret)) + { + /* Fallback to the syscall if the kernel would block. */ + int err = INTERNAL_SYSCALL_ERRNO (ret); + if (err == EAGAIN && !(flags & GRND_NONBLOCK)) + goto out; + + __set_errno (err); + ret = -1; + } + r = true; + +out: + atomic_store_relaxed (&self->getrandom_buf, state); + return r ? ret : getrandom_syscall (buffer, length, flags, cancel); +} +#endif + +/* Re-add the state state from CURP on the free list. This function is + called after fork returns in the child, so no locking is required. */ +void +__getrandom_reset_state (struct pthread *curp) +{ +#ifdef HAVE_GETRANDOM_VSYSCALL + if (grnd_alloc.states == NULL || curp->getrandom_buf == NULL) + return; + assert (grnd_alloc.len < grnd_alloc.cap); + grnd_alloc.states[grnd_alloc.len++] = release_ptr (curp->getrandom_buf); + curp->getrandom_buf = NULL; +#endif +} + +/* Called when a thread terminates, and adds its random buffer back into the + allocator pool for use in a future thread. This is called by + pthread_create during thread termination, and after signal has been + blocked. */ +void +__getrandom_vdso_release (struct pthread *curp) +{ +#ifdef HAVE_GETRANDOM_VSYSCALL + if (curp->getrandom_buf == NULL) + return; + + __libc_lock_lock (grnd_alloc.lock); + grnd_alloc.states[grnd_alloc.len++] = curp->getrandom_buf; + __libc_lock_unlock (grnd_alloc.lock); +#endif +} + +/* Reset the internal lock state in case another thread has locked while + this thread calls fork. The stale thread states will be handled by + reclaim_stacks which calls __getrandom_reset_state on each thread. */ +void +__getrandom_fork_subprocess (void) +{ +#ifdef HAVE_GETRANDOM_VSYSCALL + grnd_alloc.lock = LLL_LOCK_INITIALIZER; +#endif +} + +ssize_t +__getrandom_nocancel (void *buffer, size_t length, unsigned int flags) +{ +#ifdef HAVE_GETRANDOM_VSYSCALL + return getrandom_vdso (buffer, length, flags, false); +#else + return getrandom_syscall (buffer, length, flags, false); +#endif +} + /* Write up to LENGTH bytes of randomness starting at BUFFER. Return the number of bytes written, or -1 on error. */ ssize_t __getrandom (void *buffer, size_t length, unsigned int flags) { - return SYSCALL_CANCEL (getrandom, buffer, length, flags); +#ifdef HAVE_GETRANDOM_VSYSCALL + return getrandom_vdso (buffer, length, flags, true); +#else + return getrandom_syscall (buffer, length, flags, true); +#endif } libc_hidden_def (__getrandom) weak_alias (__getrandom, getrandom) |