Skip to contentSkip to conversation

implement getrandom() in vDSO

PATCH Superseded Jason A. Donenfeld 7 participants 37 messages Nov 29, 2022 – Dec 2, 2022

Subsystems (21)
Arm (32-bit) x86 PowerPC MIPS +17 more subsystems

v10 of 14 Superseded Latest is v14 →

lore
Share & tools
lore

Jason A. Donenfeld

author maintainer
#0 lore raw

v10 cover implement getrandom() in vDSO

Changes v9->v10:
---------------
- Actually include all 4 patches.

Changes v8->v9:
--------------
- Allocate system call number on all architectures, and split that off into a separate commit (2/4).
- Declare missing prototype in syscalls.h.

Changes v7->v8:
--------------
- Move lib/vdso/getrandom.h to include/vdso/getrandom.h in order to avoid #include "../../../../../../../../../../../......".
- Make use of two Kconfig symbols, VDSO_GETRANDOM and VGETRANDOM_ALLOC_SYSCALL, to handle selecting dependencies and conditionalizing code.
- Rename chacha20_blocks_nostack assembly function to __arch_chacha20_blocks_nostack, which allows removing the awkward C inline wrapper function.
- Save a byte per instruction by using movups instead of movdqu, and don't bother clearing registers that hold public constants.
- Add basic signal handling reentrancy protection to vDSO function.
- Invalidate RNG generation counter if key-refresh syscall fails.
- Reduce the defines in getrandom.c, which in turn requires using INT_MAX & PAGE_MASK explicitly rather than the MAX_RW_COUNT macro.
- Make use of 100 columns when it makes sense, and reformat various bits of code for clarity.
- Thoroughly document functions and add kernel-doc comments to several key functions.
- Hoist out repeated sizeof(struct vgetrandom_state) into variable.
- Rename memcpy_and_zero to memcpy_and_zero_src, and define helper macro outside of function.
- Separate all x86 work, including syscall wiring, into the x86 commit, so that the first two commits of this series are purely generic.

Changes v6->v7:
--------------
- VERY EXCITING! There is now a rudimentary glibc implementation for this from one of the glibc maintainers, Adhemerval Zanella (CC'd). A commit that works with with this latest v7 revision is here:

https://github.com/bminor/glibc/commit/247ec6dd77ec2a047163fe3a1b60e57880464b39

- Pass an unsigned int * instead of an unsigned long * for the syscall, to avoid having to add a compat syscall.
- Use ordinary function framing in assembly, rather than kernel-specific framing.
- Don't hardcode the number '2', but derive it at compile time with the expression sizeof(state->batch_key) / CHACHA_BLOCK_SIZE, as well as adding a BUILD_BUG_ON() in case that doesn't divide cleanly.

Changes v5->v6:
--------------
- Fix various build errors for odd configurations.
- Do not leak any secrets onto the stack at all, to account for possibility of fork()ing in a multithreaded scenario, which would ruin forward secrecy. Instead provide a arch-specific implementation that doesn't need stack space.
- Prevent page alignment from overflowing variable, and clamp to acceptable limits.
- Read/write unaligned bytes using get/put_unaligned.
- Add extensive comments to vDSO function explaining subtle aspects.
- Account for fork() races when writing generation counter.

Changes v4->v5:
--------------
- Add example code to vDSO addition commit showing intended use and interaction with allocations.
- Reset buffer to beginning when retrying.
- Rely on generation counter never being zero for fork detection, rather than adding extra boolean.
- Make use of __ARCH_WANT_VGETRANDOM_ALLOC macro around new syscall so that it's condition by archs that actually choose to add this, and don't forget to bump __NR_syscalls.
- Separate __cvdso_getrandom() into __cvdso_getrandom() and __cvdso_getrandom_data() so that powerpc can make a more efficient call.

Changes v3->v4:
--------------
- Split up into small series rather than one big patch.
- Use proper ordering in generation counter reads.
- Make properly generic, not just a hairball with x86, by moving symbols into correct files.

Changes v2->v3:
--------------

Big changes:

Thomas' previous objection was two-fold: 1) vgetrandom should really have the same function signature as getrandom, in addition to all of the same behavior, and 2) having vgetrandom_alloc be a vDSO function doesn't make sense, because it doesn't actually need anything from the VDSO data page and it doesn't correspond to an existing syscall.

After a discussion at Plumbers this last week, we devised the following ways to fix these: 1) we make the opque state argument be the last argument of vgetrandom, rather than the first one, since the real syscall ignores the additional argument, and that way all the registers are the same, and no behavior changes; and 2) we make vgetrandom_alloc a syscall, rather than a vDSO function, which also gives it added flexibility for the future, which is good.

Making those changes also reduced the size of this patch a bit.

Smaller changes:
- Properly add buffer offset position.
- Don't EXPORT_SYMBOL for vDSO code.
- Account for timens and vvar being in swapped pages.

--------------

Two statements:

1) Userspace wants faster cryptographically secure random numbers of arbitrary size, big or small.

2) Userspace is currently unable to safely roll its own RNG with the same security profile as getrandom().

Statement (1) has been debated for years, with arguments ranging from "we need faster cryptographically secure card shuffling!" to "the only things that actually need good randomness are keys, which are few and far between" to "actually, TLS CBC nonces are frequent" and so on. I don't intend to wade into that debate substantially, except to note that recently glibc added arc4random(), whose goal is to return a cryptographically secure uint32_t, and there are real user reports of it being too slow. So here we are.

Statement (2) is more interesting. The kernel is the nexus of all entropic inputs that influence the RNG. It is in the best position, and probably the only position, to decide anything at all about the current state of the RNG and of its entropy. One of the things it uniquely knows about is when reseeding is necessary.

For example, when a virtual machine is forked, restored, or duplicated, it's imparative that the RNG doesn't generate the same outputs. For this reason, there's a small protocol between hypervisors and the kernel that indicates this has happened, alongside some ID, which the RNG uses to immediately reseed, so as not to return the same numbers. Were userspace to expand a getrandom() seed from time T1 for the next hour, and at some point T2 < hour, the virtual machine forked, userspace would continue to provide the same numbers to two (or more) different virtual machines, resulting in potential cryptographic catastrophe. Something similar happens on resuming from hibernation (or even suspend), with various compromise scenarios there in mind.

There's a more general reason why userspace rolling its own RNG from a getrandom() seed is fraught. There's a lot of attention paid to this particular Linuxism we have of the RNG being initialized and thus non-blocking or uninitialized and thus blocking until it is initialized. These are our Two Big States that many hold to be the holy differentiating factor between safe and not safe, between cryptographically secure and garbage. The fact is, however, that the distinction between these two states is a hand-wavy wishy-washy inexact approximation. Outside of a few exceptional cases (e.g. a HW RNG is available), we actually don't really ever know with any rigor at all when the RNG is safe and ready (nor when it's compromised). We do the best we can to "estimate" it, but entropy estimation is fundamentally impossible in the general case. So really, we're just doing guess work, and hoping it's good and conservative enough. Let's then assume that there's always some potential error involved in this differentiator.

In fact, under the surface, the RNG is engineered around a different principal, and that is trying to *use* new entropic inputs regularly and at the right specific moments in time. For example, close to boot time, the RNG reseeds itself more often than later. At certain events, like VM fork, the RNG reseeds itself immediately. The various heuristics for when the RNG will use new entropy and how often is really a core aspect of what the RNG has some potential to do decently enough (and something that will probably continue to improve in the future from random.c's present set of algorithms). So in your mind, put away the metal attachment to the Two Big States, which represent an approximation with a potential margin of error. Instead keep in mind that the RNG's primary operating heuristic is how often and exactly when it's going to reseed.

So, if userspace takes a seed from getrandom() at point T1, and uses it for the next hour (or N megabytes or some other meaningless metric), during that time, potential errors in the Two Big States approximation are amplified. During that time potential reseeds are being lost, forgotten, not reflected in the output stream. That's not good.

The simplest statement you could make is that userspace RNGs that expand a getrandom() seed at some point T1 are nearly always *worse*, in some way, than just calling getrandom() every time a random number is desired.

For those reasons, after some discussion on libc-alpha, glibc's arc4random() now just calls getrandom() on each invocation. That's trivially safe, and gives us latitude to then make the safe thing faster without becoming unsafe at our leasure. Card shuffling isn't particularly fast, however.

How do we rectify this? By putting a safe implementation of getrandom() in the vDSO, which has access to whatever information a particular iteration of random.c is using to make its decisions. I use that careful language of "particular iteration of random.c", because the set of things that a vDSO getrandom() implementation might need for making decisions as good as the kernel's will likely change over time. This isn't just a matter of exporting certain *data* to userspace. We're not going to commit to a "data API" where the various heuristics used are exposed, locking in how the kernel works for decades to come, and then leave it to various userspaces to roll something on top and shoot themselves in the foot and have all sorts of complexity disasters. Rather, vDSO getrandom() is supposed to be the *same exact algorithm* that runs in the kernel, except it's been hoisted into userspace as much as possible. And so vDSO getrandom() and kernel getrandom() will always mirror each other hermetically.

API-wise, the vDSO gains this function:

  ssize_t vgetrandom(void *buffer, size_t len, unsigned int flags, void *opaque_state);

The return value and the first 3 arguments are the same as ordinary getrandom(), while the last argument is a pointer to some state allocated with vgetrandom_alloc(), explained below. Were all four arguments passed to the getrandom syscall, nothing different would happen, and the functions would have the exact same behavior.

Then, we introduce a new syscall:

void *vgetrandom_alloc([inout] unsigned int *num, [out] unsigned int *size_per_each, unsigned int flags);

This takes the desired number of opaque states in num, and returns a pointer to an array of opaque states, the number actually allocated back in num, and the size in bytes of each one in size_per_each, enabling a libc to slice up the returned array into a state per each thread. (The flags argument is always zero for now.) We very intentionally do *not* leave state allocation up to the caller of vgetrandom, but provide vgetrandom_alloc for that allocation. There are too many weird things that can go wrong, and it's important that vDSO does not provide too generic of a mechanism. It's not going to store its state in just any old memory address. It'll do it only in ones it allocates.

Right now this means it's a mlock'd page with WIPEONFORK set. In the future maybe there will be other interesting page flags or anti-heartbleed measures, or other platform-specific kernel-specific things that can be set from the syscall. Again, it's important that the kernel has a say in how this works rather than agreeing to operate on any old address; memory isn't neutral.

The syscall currently accomplishes this with a call to vm_mmap() and then a call to do_madvise(). It'd be nice to do this all at once, but I'm not sure that a helper function exists for that now, and it seems a bit premature to add one, at least for now.

The interesting meat of the implementation is in lib/vdso/getrandom.c, as generic C code, and it aims to mainly follow random.c's buffered fast key erasure logic. Before the RNG is initialized, it falls back to the syscall. Right now it uses a simple generation counter to make its decisions on reseeding (though this could be made more extensive over time).

The actual place that has the most work to do is in all of the other files. Most of the vDSO shared page infrastructure is centered around gettimeofday, and so the main structs are all in arrays for different timestamp types, and attached to time namespaces, and so forth. I've done the best I could to add onto this in an unintrusive way.

In my test results, performance is pretty stellar (around 15x for uint32_t generation), and it seems to be working. There's an extended example in the second commit of this series, showing how the syscall and the vDSO function are meant to be used together.

Cc Thomas Gleixner <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Adhemerval Zanella Netto <[email protected]>
Cc Carlos O'Donell <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Christian Brauner <[email protected]>

Jason A. Donenfeld (4):
random: add vgetrandom_alloc() syscall
arch: allocate vgetrandom_alloc() syscall number
random: introduce generic vDSO getrandom() implementation
x86: vdso: Wire up getrandom() vDSO implementation

 MAINTAINERS                                   |   2 +
 arch/alpha/kernel/syscalls/syscall.tbl        |   1 +
 arch/arm/tools/syscall.tbl                    |   1 +
 arch/arm64/include/asm/unistd.h               |   2 +-
 arch/arm64/include/asm/unistd32.h             |   2 +
 arch/ia64/kernel/syscalls/syscall.tbl         |   1 +
 arch/m68k/kernel/syscalls/syscall.tbl         |   1 +
 arch/microblaze/kernel/syscalls/syscall.tbl   |   1 +
 arch/mips/kernel/syscalls/syscall_n32.tbl     |   1 +
 arch/mips/kernel/syscalls/syscall_n64.tbl     |   1 +
 arch/mips/kernel/syscalls/syscall_o32.tbl     |   1 +
 arch/parisc/kernel/syscalls/syscall.tbl       |   1 +
 arch/powerpc/kernel/syscalls/syscall.tbl      |   1 +
 arch/s390/kernel/syscalls/syscall.tbl         |   1 +
 arch/sh/kernel/syscalls/syscall.tbl           |   1 +
 arch/sparc/kernel/syscalls/syscall.tbl        |   1 +
 arch/x86/Kconfig                              |   1 +
 arch/x86/entry/syscalls/syscall_32.tbl        |   1 +
 arch/x86/entry/syscalls/syscall_64.tbl        |   1 +
 arch/x86/entry/vdso/Makefile                  |   3 +-
 arch/x86/entry/vdso/vdso.lds.S                |   2 +
 arch/x86/entry/vdso/vgetrandom-chacha.S       | 177 +++++++++++++++
 arch/x86/entry/vdso/vgetrandom.c              |  17 ++
 arch/x86/include/asm/vdso/getrandom.h         |  55 +++++
 arch/x86/include/asm/vdso/vsyscall.h          |   2 +
 arch/x86/include/asm/vvar.h                   |  16 ++
 arch/xtensa/kernel/syscalls/syscall.tbl       |   1 +
 drivers/char/random.c                         |  84 ++++++++
 include/linux/syscalls.h                      |   3 +
 include/uapi/asm-generic/unistd.h             |   5 +-
 include/vdso/datapage.h                       |  11 +
 include/vdso/getrandom.h                      |  24 +++
 kernel/sys_ni.c                               |   3 +
 lib/vdso/Kconfig                              |  14 +-
 lib/vdso/getrandom.c                          | 204 ++++++++++++++++++
 tools/include/uapi/asm-generic/unistd.h       |   5 +-
 .../arch/mips/entry/syscalls/syscall_n64.tbl  |   1 +
 .../arch/powerpc/entry/syscalls/syscall.tbl   |   1 +
 .../perf/arch/s390/entry/syscalls/syscall.tbl |   1 +
 .../arch/x86/entry/syscalls/syscall_64.tbl    |   1 +
 40 files changed, 647 insertions(+), 5 deletions(-)
 create mode 100644 arch/x86/entry/vdso/vgetrandom-chacha.S
 create mode 100644 arch/x86/entry/vdso/vgetrandom.c
 create mode 100644 arch/x86/include/asm/vdso/getrandom.h
 create mode 100644 include/vdso/getrandom.h
 create mode 100644 lib/vdso/getrandom.c
Signature · 1 line
2.38.1
author maintainer
#1 lore raw

PATCH v10 1/4 random: add vgetrandom_alloc() syscall

The vDSO getrandom() works over an opaque per-thread state of an unexported size, which must be marked as MADV_WIPEONFORK and be mlock()'d for proper operation. Over time, the nuances of these allocations may change or grow or even differ based on architectural features.

The syscall has the signature:

void *vgetrandom_alloc([inout] unsigned int *num,
[out] unsigned int *size_per_each, unsigned int flags);

This takes the desired number of opaque states in num, and returns a pointer to an array of opaque states, the number actually allocated back in num, and the size in bytes of each one in size_per_each, enabling a libc to slice up the returned array into a state per each thread. (The flags argument is always zero for now.) Libc is expected to allocate a chunk of these on first use, and then dole them out to threads as they're created, allocating more when needed. The following commit shows an example of this, being used in conjunction with the getrandom() vDSO function.

We very intentionally do *not* leave state allocation for vDSO getrandom() up to userspace itself, but rather provide this new syscall for such allocations. vDSO getrandom() must not store its state in just any old memory address, but rather just ones that the kernel specially allocates for it, leaving the particularities of those allocations up to the kernel.

---
 MAINTAINERS              |  1 +
 drivers/char/random.c    | 75 ++++++++++++++++++++++++++++++++++++++++
 include/linux/syscalls.h |  3 ++
 include/vdso/getrandom.h | 24 +++++++++++++
 kernel/sys_ni.c          |  3 ++
 lib/vdso/Kconfig         |  7 ++++
 6 files changed, 113 insertions(+)
 create mode 100644 include/vdso/getrandom.h
MAINTAINERS +1 -0
index 256f03904987..3894f947a507 100644
@@ -17287,6 +17287,7 @@ T:	git https://git.kernel.org/pub/scm/linux/kernel/git/crng/random.git
 S:	Maintained
 F:	drivers/char/random.c
 F:	drivers/virt/vmgenid.c
+F:	include/vdso/getrandom.h
 
 RAPIDIO SUBSYSTEM
 M:	Matt Porter <[email protected]>
drivers/char/random.c +75 -0

99 lines not on this page · raw message

include/linux/syscalls.h +3 -0
index a34b0f9a9972..7741dc94f10c 100644
@@ -1006,6 +1006,9 @@ asmlinkage long sys_seccomp(unsigned int op, unsigned int flags,
 			    void __user *uargs);
 asmlinkage long sys_getrandom(char __user *buf, size_t count,
 			      unsigned int flags);
+asmlinkage long sys_vgetrandom_alloc(unsigned int __user *num,
+				     unsigned int __user *size_per_each,
+				     unsigned int flags);
 asmlinkage long sys_memfd_create(const char __user *uname_ptr, unsigned int flags);
 asmlinkage long sys_bpf(int cmd, union bpf_attr *attr, unsigned int size);
 asmlinkage long sys_execveat(int dfd, const char __user *filename,
include/vdso/getrandom.h new file +24 -0
new file mode 100644
index 000000000000..5f04c8bf4bd4
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2022 Jason A. Donenfeld <[email protected]>. All Rights Reserved.
+ */
+
+#ifndef _VDSO_GETRANDOM_H
+#define _VDSO_GETRANDOM_H
+
+#include <crypto/chacha.h>
+
+struct vgetrandom_state {
+	union {
+		struct {
+			u8 batch[CHACHA_BLOCK_SIZE * 3 / 2];
+			u32 key[CHACHA_KEY_SIZE / sizeof(u32)];
+		};
+		u8 batch_key[CHACHA_BLOCK_SIZE * 2];
+	};
+	unsigned long generation;
+	u8 pos;
+	bool in_use;
+};
+
+#endif /* _VDSO_GETRANDOM_H */
kernel/sys_ni.c +3 -0
index 860b2dcf3ac4..f28196cb919b 100644
@@ -360,6 +360,9 @@ COND_SYSCALL(pkey_free);
 /* memfd_secret */
 COND_SYSCALL(memfd_secret);
 
+/* random */
+COND_SYSCALL(vgetrandom_alloc);
+
 /*
  * Architecture specific weak syscall entries.
  */
lib/vdso/Kconfig +7 -0
index d883ac299508..b22584f8da03 100644
@@ -31,3 +31,10 @@ config GENERIC_VDSO_TIME_NS
 	  VDSO
 
 endif
+
+config VGETRANDOM_ALLOC_SYSCALL
+	bool
+	select ADVISE_SYSCALLS
+	help
+	  Selected by the getrandom() vDSO function, which requires this
+	  for state allocation.
Signature · 1 line
2.38.1
#2 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Jason!

Quoted 11 lines · Jason A. Donenfeld
drivers/char/random.c +11 -0
+
+/********************************************************************
+ *
+ * vDSO support helpers.
+ *
+ * The actual vDSO function is defined over in lib/vdso/getrandom.c,
+ * but this section contains the kernel-mode helpers to support that.
+ *
+ ********************************************************************/
+
+#ifdef CONFIG_VGETRANDOM_ALLOC_SYSCALL

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 195

Quoted 10 lines
drivers/char/random.c +10 -0
+/**
+ * vgetrandom_alloc - allocate opaque states for use with vDSO getrandom().
+ *
+ * @num: on input, a pointer to a suggested hint of how many states to
+ * allocate, and on output the number of states actually allocated.
+ *
+ * @size_per_each: the size of each state allocated, so that the caller can
+ * split up the returned allocation into individual states.
+ *
+ * @flags: currently always zero.

NIT!

I personally prefer and ask for it in stuff I maintain:

* @num: On input, a pointer to a suggested hint of how many states to
* allocate, and on output the number of states actually allocated.
*
* @size_per_each: The size of each state allocated, so that the caller can
* split up the returned allocation into individual states.
*
* @flags: Currently always zero.

But your turf :)

Quoted 7 lines

+ *
+ * The getrandom() vDSO function in userspace requires an opaque state, which
+ * this function allocates by mapping a certain number of special pages into
+ * the calling process. It takes a hint as to the number of opaque states
+ * desired, and provides the caller with the number of opaque states actually
+ * allocated, the size of each one in bytes, and the address of the first
+ * state.

make W=1 rightfully complains about:

Quoted 1 line

+

drivers/char/random.c:182: warning: bad line:

Quoted 1 line

+ * Returns a pointer to the first state in the allocation.

I have serious doubts that this statement is correct.

Think about this comment and documentation as a boiler plate for the mandatory man page for a new syscall (hint...)

Quoted 2 lines

+ *
+ */

and W=1 also complains rightfully here:

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 208

Quoted 2 lines
drivers/char/random.c +2 -0
+SYSCALL_DEFINE3(vgetrandom_alloc, unsigned int __user *, num,
+		unsigned int __user *, size_per_each, unsigned int, flags)

drivers/char/random.c:188: warning: expecting prototype for vgetrandom_alloc(). Prototype was for sys_vgetrandom_alloc() instead

Quoted 19 lines
drivers/char/random.c +1 -0
+{
include/vdso/getrandom.h new file +12 -0
new file mode 100644
index 000000000000..5f04c8bf4bd4
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2022 Jason A. Donenfeld <[email protected]>. All Rights Reserved.
+ */
+
+#ifndef _VDSO_GETRANDOM_H
+#define _VDSO_GETRANDOM_H
+
+#include <crypto/chacha.h>
+
+struct vgetrandom_state {
+	union {

include/vdso/getrandom.h line 22

Quoted 10 lines
include/vdso/getrandom.h new file +10 -0
+		struct {
+			u8 batch[CHACHA_BLOCK_SIZE * 3 / 2];
+			u32 key[CHACHA_KEY_SIZE / sizeof(u32)];
+		};
+		u8 batch_key[CHACHA_BLOCK_SIZE * 2];
+	};
+	unsigned long generation;
+	u8 pos;
+	bool in_use;
+};

Again, please make this properly tabular:

struct vgetrandom_state {
	union {
		struct {
			u8	batch[CHACHA_BLOCK_SIZE * 3 / 2];
			u32	key[CHACHA_KEY_SIZE / sizeof(u32)];
		};
		u8	batch_key[CHACHA_BLOCK_SIZE * 2];
	};
	unsigned long	generation;
	u8		pos;
	bool		in_use;
};

Plus some kernel doc which explains what this is about.

Thanks,

tglx

author maintainer
#3 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Hi Thomas,

Thanks again for the big review. Comments inline below.

Quoted 23 lines · Thomas Gleixner
Quoted 10 lines

+/**
+ * vgetrandom_alloc - allocate opaque states for use with vDSO getrandom().
+ *
+ * @num: on input, a pointer to a suggested hint of how many states to
+ * allocate, and on output the number of states actually allocated.
+ *
+ * @size_per_each: the size of each state allocated, so that the caller can
+ * split up the returned allocation into individual states.
+ *
+ * @flags: currently always zero.

NIT!

I personally prefer and ask for it in stuff I maintain:

* @num: On input, a pointer to a suggested hint of how many states to
* allocate, and on output the number of states actually allocated.
*
* @size_per_each: The size of each state allocated, so that the caller can
* split up the returned allocation into individual states.
*
* @flags: Currently always zero.

Quoted 1 line

But your turf :)

Hm. Caps and punctuation seem mostly missing in kernel/time/, though it is that way in some places, so I'll do it with caps and punctuation. Presumably that's the "newer" style you prefer, though I didn't look at the dates in git-blame to confirm that supposition.

Quoted 11 lines
Quoted 8 lines

+ *
+ * The getrandom() vDSO function in userspace requires an opaque state, which
+ * this function allocates by mapping a certain number of special pages into
+ * the calling process. It takes a hint as to the number of opaque states
+ * desired, and provides the caller with the number of opaque states actually
+ * allocated, the size of each one in bytes, and the address of the first
+ * state.

make W=1 rightfully complains about:

Quoted 7 lines
Quoted 1 line

+

drivers/char/random.c:182: warning: bad line:

Quoted 1 line

+ * Returns a pointer to the first state in the allocation.

I have serious doubts that this statement is correct.

"Returns the address of the first state in the allocation" is better I guess.

Quoted 6 lines

and W=1 also complains rightfully here:

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 208

Quoted 2 lines
drivers/char/random.c +2 -0
+SYSCALL_DEFINE3(vgetrandom_alloc, unsigned int __user *, num,
+		unsigned int __user *, size_per_each, unsigned int, flags)

drivers/char/random.c:188: warning: expecting prototype for vgetrandom_alloc(). Prototype was for sys_vgetrandom_alloc() instead

Squinted at a lot of headers before realizing that's a kernel-doc warning. Fixed, thanks.

Quoted 33 lines
Quoted 7 lines
include/vdso/getrandom.h new file +7 -0
+#ifndef _VDSO_GETRANDOM_H
+#define _VDSO_GETRANDOM_H
+
+#include <crypto/chacha.h>
+
+struct vgetrandom_state {
+	union {

include/vdso/getrandom.h line 22

Quoted 10 lines
include/vdso/getrandom.h new file +10 -0
+		struct {
+			u8 batch[CHACHA_BLOCK_SIZE * 3 / 2];
+			u32 key[CHACHA_KEY_SIZE / sizeof(u32)];
+		};
+		u8 batch_key[CHACHA_BLOCK_SIZE * 2];
+	};
+	unsigned long generation;
+	u8 pos;
+	bool in_use;
+};

Again, please make this properly tabular:

struct vgetrandom_state {
	union {
		struct {
			u8	batch[CHACHA_BLOCK_SIZE * 3 / 2];
			u32	key[CHACHA_KEY_SIZE / sizeof(u32)];
		};
		u8	batch_key[CHACHA_BLOCK_SIZE * 2];
	};
	unsigned long	generation;
	u8		pos;
	bool		in_use;
};
Quoted 1 line

Plus some kernel doc which explains what this is about.

Will do. Though, I'm going to move this to the vDSO commit, and for the syscall commit, which needs the struct to merely exist, I'll have no members in it. This should make review a bit easier.

Jason

#4 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Quoted 7 lines · Jason A. Donenfeld
Quoted 3 lines · Thomas Gleixner
Quoted 1 line

+ * Returns a pointer to the first state in the allocation.

I have serious doubts that this statement is correct.

"Returns the address of the first state in the allocation" is better I guess.

Does not even come close to correct.

As my previous hint of 'using this as template for the (hint:missing) man page' did not work well, may I suggest that you look at the various return statements in that function and validate whether your proposed return value documentation is valid for all of them?

Thanks,

tglx

author maintainer
#5 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Quoted 9 lines · Thomas Gleixner
Quoted 7 lines · Jason A. Donenfeld
Quoted 3 lines · Thomas Gleixner
Quoted 1 line

+ * Returns a pointer to the first state in the allocation.

I have serious doubts that this statement is correct.

"Returns the address of the first state in the allocation" is better I guess.

Quoted 6 lines

Does not even come close to correct.

As my previous hint of 'using this as template for the (hint:missing) man page' did not work well, may I suggest that you look at the various return statements in that function and validate whether your proposed return value documentation is valid for all of them?

Ahh, the error values and such. Righto. Will do. I'll match the style of similar functions.

Jason

#6 lore raw

RE: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

In reply to Thomas Gleixner

From: Thomas Gleixner

Quoted 118 lines

Sent: 29 November 2022 22:02

Jason!

Quoted 11 lines · Jason A. Donenfeld
drivers/char/random.c +11 -0
+
+/********************************************************************
+ *
+ * vDSO support helpers.
+ *
+ * The actual vDSO function is defined over in lib/vdso/getrandom.c,
+ * but this section contains the kernel-mode helpers to support that.
+ *
+ ********************************************************************/
+
+#ifdef CONFIG_VGETRANDOM_ALLOC_SYSCALL

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 195

Quoted 10 lines
drivers/char/random.c +10 -0
+/**
+ * vgetrandom_alloc - allocate opaque states for use with vDSO getrandom().
+ *
+ * @num: on input, a pointer to a suggested hint of how many states to
+ * allocate, and on output the number of states actually allocated.
+ *
+ * @size_per_each: the size of each state allocated, so that the caller can
+ * split up the returned allocation into individual states.
+ *
+ * @flags: currently always zero.

NIT!

I personally prefer and ask for it in stuff I maintain:

* @num: On input, a pointer to a suggested hint of how many states to
* allocate, and on output the number of states actually allocated.
*
* @size_per_each: The size of each state allocated, so that the caller can
* split up the returned allocation into individual states.
*
* @flags: Currently always zero.

But your turf :)

Quoted 7 lines

+ *
+ * The getrandom() vDSO function in userspace requires an opaque state, which
+ * this function allocates by mapping a certain number of special pages into
+ * the calling process. It takes a hint as to the number of opaque states
+ * desired, and provides the caller with the number of opaque states actually
+ * allocated, the size of each one in bytes, and the address of the first
+ * state.

make W=1 rightfully complains about:

Quoted 1 line

+

drivers/char/random.c:182: warning: bad line:

Quoted 1 line

+ * Returns a pointer to the first state in the allocation.

I have serious doubts that this statement is correct.

Think about this comment and documentation as a boiler plate for the mandatory man page for a new syscall (hint...)

Quoted 2 lines

+ *
+ */

and W=1 also complains rightfully here:

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 208

Quoted 2 lines
drivers/char/random.c +2 -0
+SYSCALL_DEFINE3(vgetrandom_alloc, unsigned int __user *, num,
+		unsigned int __user *, size_per_each, unsigned int, flags)

drivers/char/random.c:188: warning: expecting prototype for vgetrandom_alloc(). Prototype was for sys_vgetrandom_alloc() instead

Quoted 19 lines
drivers/char/random.c +1 -0
+{
include/vdso/getrandom.h new file +12 -0
new file mode 100644
index 000000000000..5f04c8bf4bd4
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2022 Jason A. Donenfeld <[email protected]>. All Rights Reserved.
+ */
+
+#ifndef _VDSO_GETRANDOM_H
+#define _VDSO_GETRANDOM_H
+
+#include <crypto/chacha.h>
+
+struct vgetrandom_state {
+	union {

include/vdso/getrandom.h line 22

Quoted 10 lines
include/vdso/getrandom.h new file +10 -0
+		struct {
+			u8 batch[CHACHA_BLOCK_SIZE * 3 / 2];
+			u32 key[CHACHA_KEY_SIZE / sizeof(u32)];
+		};
+		u8 batch_key[CHACHA_BLOCK_SIZE * 2];
+	};
+	unsigned long generation;
+	u8 pos;
+	bool in_use;
+};

Again, please make this properly tabular:

struct vgetrandom_state {
	union {
		struct {
			u8	batch[CHACHA_BLOCK_SIZE * 3 / 2];
			u32	key[CHACHA_KEY_SIZE / sizeof(u32)];
		};
		u8	batch_key[CHACHA_BLOCK_SIZE * 2];
	};
	unsigned long	generation;
	u8		pos;
	bool		in_use;
};
Quoted 1 line

Plus some kernel doc which explains what this is about.

That structure looks horrid - especially for something shared between entities.
The 'unsigned long' should be either u32 or u64. There is 'hidden' padding at the end.
If 'pos' is an index into something (longer name would be better) then there is no reason to squeeze the value into 1 byte - it doesn't save anything and might make things bigger.

(I think Jason might have blocked my emails, he doesn't like critisicism/feedback.)

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)

author maintainer
#7 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Quoted 13 lines · David Laight
Quoted 13 lines
struct vgetrandom_state {
	union {
		struct {
			u8	batch[CHACHA_BLOCK_SIZE * 3 / 2];
			u32	key[CHACHA_KEY_SIZE / sizeof(u32)];
		};
		u8	batch_key[CHACHA_BLOCK_SIZE * 2];
	};
	unsigned long	generation;
	u8		pos;
	bool		in_use;
};
Quoted 5 lines
Quoted 1 line

Plus some kernel doc which explains what this is about.

That structure looks horrid - especially for something shared between entities.
The 'unsigned long' should be either u32 or u64.

This struct isn't shared. It's used only by user mode code. There may well be other issues with that long, though.

Jason

#8 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

In reply to Jason A. Donenfeld

* Jason A. Donenfeld:

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 190

Quoted 6 lines
drivers/char/random.c +6 -0
+#ifdef CONFIG_VGETRANDOM_ALLOC_SYSCALL
+/**
+ * vgetrandom_alloc - allocate opaque states for use with vDSO getrandom().
+ *
+ * @num: on input, a pointer to a suggested hint of how many states to
+ * allocate, and on output the number of states actually allocated.

Should userspace call this system call again if it needs more states? The interface description doesn't make this clear.

Quoted 7 lines

+ * @size_per_each: the size of each state allocated, so that the caller can
+ * split up the returned allocation into individual states.
+ *
+ * @flags: currently always zero.
+ *
+ * The getrandom() vDSO function in userspace requires an opaque state, which
+ * this function allocates by mapping a certain number of special pages into

Quoted 8 lines

+ * the calling process. It takes a hint as to the number of opaque states
+ * desired, and provides the caller with the number of opaque states actually
+ * allocated, the size of each one in bytes, and the address of the first
+ * state.
+
+ * Returns a pointer to the first state in the allocation.
+ *
+ */

How do we deallocate this memory? Must it remain permanently allocated?

Can userspace use the memory for something else if it's not passed to getrandom? The separate system call strongly suggests that the allocation is completely owned by the kernel, but there isn't documentation here how the allocation life-cycle is supposed to look like. In particular, it is not clear if vgetrandom_alloc or getrandom could retain a reference to the allocation in a future implementation of these interfaces.

Some users might want to zap the memory for extra hardening after use, and it's not clear if that's allowed, either.

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 209

Quoted 3 lines
drivers/char/random.c +3 -0
+SYSCALL_DEFINE3(vgetrandom_alloc, unsigned int __user *, num,
+		unsigned int __user *, size_per_each, unsigned int, flags)
+{

ABI-wise, that should work.

Quoted 4 lines
drivers/char/random.c +4 -0
+	const size_t state_size = sizeof(struct vgetrandom_state);
+	size_t alloc_size, num_states;
+	unsigned long pages_addr;
+	unsigned int num_hint;

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 223

Quoted 10 lines
drivers/char/random.c +10 -0
+	int ret;
+
+	if (flags)
+		return -EINVAL;
+
+	if (get_user(num_hint, num))
+		return -EFAULT;
+
+	num_states = clamp_t(size_t, num_hint, 1, (SIZE_MAX & PAGE_MASK) / state_size);
+	alloc_size = PAGE_ALIGN(num_states * state_size);

Doesn't this waste space for one state if state_size happens to be a power of 2? Why do this SIZE_MAX & PAGE_MASK thing at all? Shouldn't it be PAGE_SIZE / state_size?

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 229

Quoted 5 lines
drivers/char/random.c +5 -0
+	if (put_user(alloc_size / state_size, num) || put_user(state_size, size_per_each))
+		return -EFAULT;
+
+	pages_addr = vm_mmap(NULL, 0, alloc_size, PROT_READ | PROT_WRITE,
+			     MAP_PRIVATE | MAP_ANONYMOUS | MAP_LOCKED, 0);

I think Rasmus has already raised questions about MAP_LOCKED.

I think the kernel cannot rely on it because userspace could call munlock on the allocation.

Quoted 4 lines
drivers/char/random.c +4 -0
+	if (IS_ERR_VALUE(pages_addr))
+		return pages_addr;
+
+	ret = do_madvise(current->mm, pages_addr, alloc_size, MADV_WIPEONFORK);

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 243

Quoted 10 lines
drivers/char/random.c +10 -0
+	if (ret < 0)
+		goto err_unmap;
+
+	return pages_addr;
+
+err_unmap:
+	vm_munmap(pages_addr, alloc_size);
+	return ret;
+}
+#endif

If there's no registration of the allocation, it's not clear why we need a separate system call for this. From a documentation perspective, it may be easier to describe proper use of the getrandom vDSO call if ownership resides with userspace. But it will constrain future evolution of the implementation because you can't add registration (retaining a reference to the passed-in area in getrandom) after the fact. But I'm not sure if this is possible with the current interface, either. Userspace has to make some assumptions about the life-cycle to avoid a memory leak on thread exit.

Thanks,
Florian

author maintainer
#9 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Hi Florian,

Quoted 11 lines · Florian Weimer

* Jason A. Donenfeld:

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 190

Quoted 6 lines
drivers/char/random.c +6 -0
+#ifdef CONFIG_VGETRANDOM_ALLOC_SYSCALL
+/**
+ * vgetrandom_alloc - allocate opaque states for use with vDSO getrandom().
+ *
+ * @num: on input, a pointer to a suggested hint of how many states to
+ * allocate, and on output the number of states actually allocated.

Should userspace call this system call again if it needs more states? The interface description doesn't make this clear.

Yes. And indeed that's what Adhemerval's patch does.

Quoted 17 lines
Quoted 16 lines

+ * @size_per_each: the size of each state allocated, so that the caller can
+ * split up the returned allocation into individual states.
+ *
+ * @flags: currently always zero.
+ *
+ * The getrandom() vDSO function in userspace requires an opaque state, which
+ * this function allocates by mapping a certain number of special pages into
+ * the calling process. It takes a hint as to the number of opaque states
+ * desired, and provides the caller with the number of opaque states actually
+ * allocated, the size of each one in bytes, and the address of the first
+ * state.
+
+ * Returns a pointer to the first state in the allocation.
+ *
+ */

Quoted 1 line

How do we deallocate this memory? Must it remain permanently allocated?

It can be deallocated with munmap.

Quoted 2 lines

Can userspace use the memory for something else if it's not passed to getrandom?

I suspect the documentation answer here is, "no", even if technically it might happen to work on this kernel or that kernel. I suppose this could even be quasi-enforced by xoring the top bits with some vdso compile-time constant, so you can't rely on being able to dereference it yourself.

Quoted 9 lines

The separate system call strongly suggests that the allocation is completely owned by the kernel, but there isn't documentation here how the allocation life-cycle is supposed to look like. In particular, it is not clear if vgetrandom_alloc or getrandom could retain a reference to the allocation in a future implementation of these interfaces.

Some users might want to zap the memory for extra hardening after use, and it's not clear if that's allowed, either.

I don't think zapping that memory is supported, or even a sensible thing to do. In the first place, I don't think we should suggest that the user can dereference that pointer, at all. In that sense, maybe it's best to call it a "handle" or something similar (a "HANDLE"! a "HWND"? a "HRNG"? just kidding). In the second place, the fast erasure aspect of this means that such hardening would have no effect -- the key is overwritten after using for forward secrecy, anyway, and batched bytes are zeroed. (There is a corner case that might make it interesting to wipe in the parent, not just the child, on fork, but that's sort of a separate matter and would ideally be handled by kernel space anyway.)

Quoted 4 lines

If there's no registration of the allocation, it's not clear why we need a separate system call for this. From a documentation perspective, it may be easier to describe proper use of the getrandom vDSO call if ownership resides with userspace.

No, absolutely not, for the multiple reasons already listed in the commit messages and cover letter and previous emails. But you seem aware of this:

Quoted 6 lines

But it will constrain future
evolution of the implementation because you can't add registration (retaining a reference to the passed-in area in getrandom) after the fact. But I'm not sure if this is possible with the current interface, either. Userspace has to make some assumptions about the life-cycle to avoid a memory leak on thread exit.

It sounds like this is sort of a different angle on Rasmus' earlier comment about how munmap leaks implementation details. Maybe there's something to that after all? Or not? I see two approaches:

1) Keep munmap as the allocation function. If later on we do fancy registration and in-kernel state tracking, or add fancy protection flags, or whatever else, munmap should be able to identify these pages and carry out whatever special treatment is necessary.

2) Convert vgetrandom_alloc() into a clone3-style syscall, as Christian suggested earlier, which might allow for a bit more overloading capability. That would be a struct that looks like:

      struct vgetrandom_alloc_args {
	  __aligned_u64 flags;
          __aligned_u64 states;
	  __aligned_u64 num;
	  __aligned_u64 size_of_each;
      }

- If flags is VGRA_ALLOCATE, states and size_of_each must be zero on input, while num is the hint, as is the case now. On output, states, size_of_each, and num are filled in.

- If flags is VGRA_DEALLOCATE, states, size_of_each, and num must be as they were originally, and then it deallocates.

I suppose (2) would alleviate your concerns entirely, without future uncertainty over what it'd be like to add special cases to munmap(). And it'd add a bit more future proofing to the syscall, depending on what we do.

So maybe I'm warming up to that approach a bit.

Quoted 6 lines

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 223

Quoted 2 lines
drivers/char/random.c +2 -0
+	num_states = clamp_t(size_t, num_hint, 1, (SIZE_MAX & PAGE_MASK) / state_size);
+	alloc_size = PAGE_ALIGN(num_states * state_size);

Doesn't this waste space for one state if state_size happens to be a power of 2? Why do this SIZE_MAX & PAGE_MASK thing at all? Shouldn't it be PAGE_SIZE / state_size?

The first line is a clamp. That fixes num_hint between 1 and the largest number that when multiplied and rounded up won't overflow.

So, if state_size is a power of two, let's say 256, and there's only one state, here's what that looks like:

    num_states = clamp(1, 1, (0xffffffff & (~(4096 - 1))) / 256 = 16777200) = 1
    alloc_size = PAGE_ALIGN(1 * 256) = 4096

So that seems like it's working as intended, right? Or if not, maybe it'd help to write out the digits you're concerned about?

Quoted 10 lines

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 229

Quoted 5 lines
drivers/char/random.c +5 -0
+	if (put_user(alloc_size / state_size, num) || put_user(state_size, size_per_each))
+		return -EFAULT;
+
+	pages_addr = vm_mmap(NULL, 0, alloc_size, PROT_READ | PROT_WRITE,
+			     MAP_PRIVATE | MAP_ANONYMOUS | MAP_LOCKED, 0);

I think Rasmus has already raised questions about MAP_LOCKED.

I think the kernel cannot rely on it because userspace could call munlock on the allocation.

Then they're caught holding the bag? This doesn't seem much different from userspace shooting themselves in general, like writing garbage into the allocated states and then trying to use them. If this is something you really, really are concerned about, then maybe my cheesy dumb xor thing mentioned above would be a low effort mitigation here.

Jason

author maintainer
#10 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Quoted 18 lines · Jason A. Donenfeld

2) Convert vgetrandom_alloc() into a clone3-style syscall, as Christian suggested earlier, which might allow for a bit more overloading capability. That would be a struct that looks like:

      struct vgetrandom_alloc_args {
	  __aligned_u64 flags;
          __aligned_u64 states;
	  __aligned_u64 num;
	  __aligned_u64 size_of_each;
      }

- If flags is VGRA_ALLOCATE, states and size_of_each must be zero on input, while num is the hint, as is the case now. On output, states, size_of_each, and num are filled in.

- If flags is VGRA_DEALLOCATE, states, size_of_each, and num must be as they were originally, and then it deallocates.

Quoted 6 lines

I suppose (2) would alleviate your concerns entirely, without future uncertainty over what it'd be like to add special cases to munmap(). And it'd add a bit more future proofing to the syscall, depending on what we do.

So maybe I'm warming up to that approach a bit.

So I just did a little quick implementation to see what it'd feel like, and actually, it's quite simple, and might address a lot of concerns all at once. What do you think of the below? Documentation and such still needs work obviously, but the bones should be there.

drivers/char/random.c +37 -21
index 4341c6a91207..dae6095b937d 100644
@@ -189,44 +189,53 @@ int __cold execute_with_initialized_rng(struct notifier_block *nb)
 /**
  * sys_vgetrandom_alloc - Allocate opaque states for use with vDSO getrandom().
  *
- * @num:	   On input, a pointer to a suggested hint of how many states to
- * 		   allocate, and on output the number of states actually allocated.
- *
- * @size_per_each: The size of each state allocated, so that the caller can
- *		   split up the returned allocation into individual states.
- *
- * @flags:	   Currently always zero.
+ * @uargs:	A vgetrandom_alloc_args which may be updated on return.
+ * 		allocate, and on output the number of states actually allocated.
+ * @usize:	The size of @uargs, which determines the version of the struct used.
  *
  * The getrandom() vDSO function in userspace requires an opaque state, which
  * this function allocates by mapping a certain number of special pages into
  * the calling process. It takes a hint as to the number of opaque states
  * desired, and provides the caller with the number of opaque states actually
  * allocated, the size of each one in bytes, and the address of the first
- * state.
+ * state. Alternatively, if the VGRA_DEALLOCATE flag is specified, the provided
+ * states parameter is unmapped.
  *
- * Returns the address of the first state in the allocation on success, or a
- * negative error value on failure.
+ * Returns 0 on success and an error value otherwise.
  */
-SYSCALL_DEFINE3(vgetrandom_alloc, unsigned int __user *, num,
-		unsigned int __user *, size_per_each, unsigned int, flags)
+SYSCALL_DEFINE2(vgetrandom_alloc, struct vgetrandom_alloc_args __user *, uargs, size_t, usize)
 {
 	const size_t state_size = sizeof(struct vgetrandom_state);
+	const size_t max_states = (SIZE_MAX & PAGE_MASK) / state_size;
+	struct vgetrandom_alloc_args args;
 	size_t alloc_size, num_states;
 	unsigned long pages_addr;
-	unsigned int num_hint;
 	int ret;
 
-	if (flags)
+	if (usize > PAGE_SIZE)
+		return -E2BIG;
+	if (usize < VGETRANDOM_ALLOC_ARGS_SIZE_VER0)
 		return -EINVAL;
+	ret = copy_struct_from_user(&args, sizeof(args), uargs, usize);
+	if (ret)
+		return ret;
 
-	if (get_user(num_hint, num))
-		return -EFAULT;
+	/* Currently only VGRA_DEALLOCATE is defined. */
+	if (args.flags & ~VGRA_DEALLOCATE)
+		return -EINVAL;
 
-	num_states = clamp_t(size_t, num_hint, 1, (SIZE_MAX & PAGE_MASK) / state_size);
-	alloc_size = PAGE_ALIGN(num_states * state_size);
+	if (args.flags & VGRA_DEALLOCATE) {
+		if (args.size_per_each != state_size || args.num > max_states || !args.states)
+			return -EINVAL;
+		return vm_munmap(args.states, args.num * state_size);
+	}
 
-	if (put_user(alloc_size / state_size, num) || put_user(state_size, size_per_each))
-		return -EFAULT;
+	/* These don't make sense as input values if allocating, so reject them. */
+	if (args.size_per_each || args.states)
+		return -EINVAL;
+
+	num_states = clamp_t(size_t, args.num, 1, max_states);
+	alloc_size = PAGE_ALIGN(num_states * state_size);
 
 	pages_addr = vm_mmap(NULL, 0, alloc_size, PROT_READ | PROT_WRITE,
 			     MAP_PRIVATE | MAP_ANONYMOUS | MAP_LOCKED, 0);
@@ -237,7 +246,14 @@ SYSCALL_DEFINE3(vgetrandom_alloc, unsigned int __user *, num,
 	if (ret < 0)
 		goto err_unmap;
 
-	return pages_addr;
+	args.num = num_states;
+	args.size_per_each = state_size;
+	args.states = pages_addr;
+
+	ret = -EFAULT;
+	if (copy_to_user(uargs, &args, sizeof(args)))
+		goto err_unmap;
+	return 0;
 
 err_unmap:
 	vm_munmap(pages_addr, alloc_size);
include/linux/syscalls.h +3 -3
index 7741dc94f10c..de4338e26db0 100644
@@ -72,6 +72,7 @@ struct open_how;
 struct mount_attr;
 struct landlock_ruleset_attr;
 enum landlock_rule_type;
+struct vgetrandom_alloc_args;
 
 #include <linux/types.h>
 #include <linux/aio_abi.h>
@@ -1006,9 +1007,8 @@ asmlinkage long sys_seccomp(unsigned int op, unsigned int flags,
 			    void __user *uargs);
 asmlinkage long sys_getrandom(char __user *buf, size_t count,
 			      unsigned int flags);
-asmlinkage long sys_vgetrandom_alloc(unsigned int __user *num,
-				     unsigned int __user *size_per_each,
-				     unsigned int flags);
+asmlinkage long sys_vgetrandom_alloc(struct vgetrandom_alloc_args __user *uargs,
+				     size_t size);
 asmlinkage long sys_memfd_create(const char __user *uname_ptr, unsigned int flags);
 asmlinkage long sys_bpf(int cmd, union bpf_attr *attr, unsigned int size);
 asmlinkage long sys_execveat(int dfd, const char __user *filename,
include/uapi/linux/random.h +26 -0
index e744c23582eb..49911ea2c343 100644
@@ -55,4 +55,30 @@ struct rand_pool_info {
 #define GRND_RANDOM	0x0002
 #define GRND_INSECURE	0x0004
 
+/*
+ * Flags for vgetrandom_alloc(2)
+ *
+ * VGRA_DEALLOCATE	Deallocate supplied states.
+ */
+#define VGRA_DEALLOCATE	0x0001ULL
+
+/**
+ * struct vgetrandom_alloc_args - Arguments for the vgetrandom_alloc(2) syscall.
+ *
+ * @flags:	   Zero or more VGRA_* flags.
+ * @states:	   Zero on input if allocating, and filled in on successful
+ *		   return. An existing allocation, if deallocating.
+ * @num:	   A hint as to the desired number of states, if allocating. The
+ *		   number of existing states in @states, if deallocating
+ * @size_per_each: The size of each state in @states.
+ */
+struct vgetrandom_alloc_args {
+	__aligned_u64 flags;
+	__aligned_u64 states;
+	__aligned_u64 num;
+	__aligned_u64 size_per_each;
+};
+
+#define VGETRANDOM_ALLOC_ARGS_SIZE_VER0 32 /* sizeof first published struct */
+
 #endif /* _UAPI_LINUX_RANDOM_H */
author maintainer
#11 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Quoted 24 lines · Jason A. Donenfeld
Quoted 23 lines · Jason A. Donenfeld

2) Convert vgetrandom_alloc() into a clone3-style syscall, as Christian suggested earlier, which might allow for a bit more overloading capability. That would be a struct that looks like:

      struct vgetrandom_alloc_args {
	  __aligned_u64 flags;
          __aligned_u64 states;
	  __aligned_u64 num;
	  __aligned_u64 size_of_each;
      }

- If flags is VGRA_ALLOCATE, states and size_of_each must be zero on input, while num is the hint, as is the case now. On output, states, size_of_each, and num are filled in.

- If flags is VGRA_DEALLOCATE, states, size_of_each, and num must be as they were originally, and then it deallocates.

I suppose (2) would alleviate your concerns entirely, without future uncertainty over what it'd be like to add special cases to munmap(). And it'd add a bit more future proofing to the syscall, depending on what we do.

Quoted 6 lines
Quoted 1 line

So maybe I'm warming up to that approach a bit.

So I just did a little quick implementation to see what it'd feel like, and actually, it's quite simple, and might address a lot of concerns all at once. What do you think of the below? Documentation and such still needs work obviously, but the bones should be there.

Well, despite writing into the ether here, I continue to chase my tail around in circles over this. After Adhemerval expressed a sort of "meh" opinion to me on IRC around doing the clone3-like thing, I went down a mm rabbit hole and started looking at all the various ways memory is allocated in userspace and under what conditions and for what and why. Turns out there are a few drivers doing interesting things in this space.

The long and short of it is that:
- All addresses involve maps and page tables.
- Allocating is mapping, deallocating is unmapping, and there's no way around that.
- Memory that's "special" usually comes with special attributes or operations on its vma.

So, this makes me think that munmap is the fine *and correct* API for deallocation. It's what everything else uses, even "special" things. And it doesn't constrain us in the future in case this gets "registered" somehow, as Florian described it, because it's still attached to current->mm and will still always go through the same mapping APIs anyway.

In light of that, I'm going to stick with the original API design, and not do the clone3() args struct thing and the VGRA_DEALLOCATE flag. However, I think it'd be a good idea to add an additional parameter of "unsigned long addr", which is enforced/reserved to be always 0 for now. This might prove useful for something together with the currently unused flags argument, sometime in the future.

Jason

author maintainer
#12 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

In reply to Jason A. Donenfeld
Quoted 14 lines · Jason A. Donenfeld
Quoted 2 lines

Can userspace use the memory for something else if it's not passed to getrandom?

I suspect the documentation answer here is, "no", even if technically it might happen to work on this kernel or that kernel. I suppose this could even be quasi-enforced by xoring the top bits with some vdso compile-time constant, so you can't rely on being able to dereference it yourself.
[...]
Then they're caught holding the bag? This doesn't seem much different from userspace shooting themselves in general, like writing garbage into the allocated states and then trying to use them. If this is something you really, really are concerned about, then maybe my cheesy dumb xor thing mentioned above would be a low effort mitigation here.

I implemented a sample of this, below. I think this is a bit silly, though, and making this fully robust could take some effort. Overall, I don't think we should do this.

However, the more I think about the args thing from the last email, the more I like *that* idea. So I think I'll roll with that.

But this cheesy pointer obfuscation thing here, meh. But here's what it could look like anyway:

drivers/char/random.c +2 -2
index 2aaeb48d11be..7aff45165ce5 100644
@@ -228,7 +228,7 @@ SYSCALL_DEFINE2(vgetrandom_alloc, struct vgetrandom_alloc_args __user *, uargs,
 	if (args.flags & VGRA_DEALLOCATE) {
 		if (args.size_per_each != state_size || args.num > max_states || !args.states)
 			return -EINVAL;
-		return vm_munmap(args.states, args.num * state_size);
+		return vm_munmap(args.states ^ VGETRANDOM_STATE_HI_TAINT, args.num * state_size);
 	}
 
 	/* These don't make sense as input values if allocating, so reject them. */
@@ -249,7 +249,7 @@ SYSCALL_DEFINE2(vgetrandom_alloc, struct vgetrandom_alloc_args __user *, uargs,
 
 	args.num = num_states;
 	args.size_per_each = state_size;
-	args.states = pages_addr;
+	args.states = pages_addr ^ VGETRANDOM_STATE_HI_TAINT;
 
 	ret = -EFAULT;
 	if (copy_to_user(uargs, &args, sizeof(args)))
include/vdso/getrandom.h +7 -0
index cb624799a8e7..9a6aaf4d99d4 100644
@@ -8,6 +8,7 @@
 
 #include <crypto/chacha.h>
 #include <vdso/limits.h>
+#include <linux/version.h>
 
 /**
  * struct vgetrandom_state - State used by vDSO getrandom() and allocated by vgetrandom_alloc().
@@ -41,4 +42,10 @@ struct vgetrandom_state {
 	bool 			in_use;
 };
 
+/* Be annoying by changing frequently enough. */
+#define VGETRANDOM_STATE_HI_TAINT ((unsigned long)(((LINUX_VERSION_CODE >> 16) + \
+				    (LINUX_VERSION_CODE >>  8) + (LINUX_VERSION_CODE >>  0) + \
+				    __GNUC__ + __GNUC_MINOR__ + __GNUC_PATCHLEVEL__) \
+				   & 0xff) << (BITS_PER_LONG - 8))
+
 #endif /* _VDSO_GETRANDOM_H */
lib/vdso/getrandom.c +1 -1
index 9ca624756432..14cbd349186c 100644
@@ -57,7 +57,7 @@ __cvdso_getrandom_data(const struct vdso_rng_data *rng_info, void *buffer, size_
 		       unsigned int flags, void *opaque_state)
 {
 	ssize_t ret = min_t(size_t, INT_MAX & PAGE_MASK /* = MAX_RW_COUNT */, len);
-	struct vgetrandom_state *state = opaque_state;
+	struct vgetrandom_state *state = (void *)((unsigned long)opaque_state ^ VGETRANDOM_STATE_HI_TAINT);
 	size_t batch_len, nblocks, orig_len = len;
 	unsigned long current_generation;
 	void *orig_buffer = buffer;
#13 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

In reply to Jason A. Donenfeld

* Jason A. Donenfeld:

Quoted 4 lines

I don't think zapping that memory is supported, or even a sensible thing to do. In the first place, I don't think we should suggest that the user can dereference that pointer, at all. In that sense, maybe it's best to call it a "handle" or something similar (a "HANDLE"! a "HWND"? a "HRNG"?

Surely the caller has to carve up the allocation, so the returned pointer is not opaque at all. From Adhemerval's glibc patch:

      grnd_allocator.cap = new_cap;
      grnd_allocator.states = new_states;

      for (size_t i = 0; i < num; ++i)
	{
	  grnd_allocator.states[i] = new_block;
	  new_block += size_per_each;
	}
      grnd_allocator.len = num;
    }

That's the opposite of a handle, really.

Quoted 7 lines
Quoted 6 lines

But it will constrain future
evolution of the implementation because you can't add registration (retaining a reference to the passed-in area in getrandom) after the fact. But I'm not sure if this is possible with the current interface, either. Userspace has to make some assumptions about the life-cycle to avoid a memory leak on thread exit.

Quoted 8 lines

It sounds like this is sort of a different angle on Rasmus' earlier comment about how munmap leaks implementation details. Maybe there's something to that after all? Or not? I see two approaches:

1) Keep munmap as the allocation function. If later on we do fancy registration and in-kernel state tracking, or add fancy protection flags, or whatever else, munmap should be able to identify these pages and carry out whatever special treatment is necessary.

munmap is fine, but the interface needs to say how to use it, and what length to pass.

Quoted 10 lines
Quoted 6 lines

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 223

Quoted 2 lines
drivers/char/random.c +2 -0
+	num_states = clamp_t(size_t, num_hint, 1, (SIZE_MAX & PAGE_MASK) / state_size);
+	alloc_size = PAGE_ALIGN(num_states * state_size);

Doesn't this waste space for one state if state_size happens to be a power of 2? Why do this SIZE_MAX & PAGE_MASK thing at all? Shouldn't it be PAGE_SIZE / state_size?

The first line is a clamp. That fixes num_hint between 1 and the largest number that when multiplied and rounded up won't overflow.

Quoted 8 lines

So, if state_size is a power of two, let's say 256, and there's only one state, here's what that looks like:

    num_states = clamp(1, 1, (0xffffffff & (~(4096 - 1))) / 256 = 16777200) = 1
    alloc_size = PAGE_ALIGN(1 * 256) = 4096

So that seems like it's working as intended, right? Or if not, maybe it'd help to write out the digits you're concerned about?

I think I was just confused.

Quoted 8 lines
Quoted 8 lines

drivers/char/random.c int __cold execute_with_initialized_rng(struct notifier_block *nb) · line 229

Quoted 5 lines
drivers/char/random.c +5 -0
+	if (put_user(alloc_size / state_size, num) || put_user(state_size, size_per_each))
+		return -EFAULT;
+
+	pages_addr = vm_mmap(NULL, 0, alloc_size, PROT_READ | PROT_WRITE,
+			     MAP_PRIVATE | MAP_ANONYMOUS | MAP_LOCKED, 0);

I think Rasmus has already raised questions about MAP_LOCKED.

Quoted 8 lines
Quoted 2 lines

I think the kernel cannot rely on it because userspace could call munlock on the allocation.

Then they're caught holding the bag? This doesn't seem much different from userspace shooting themselves in general, like writing garbage into the allocated states and then trying to use them. If this is something you really, really are concerned about, then maybe my cheesy dumb xor thing mentioned above would be a low effort mitigation here.

So the MAP_LOCKED is just there to prevent leakage to swap?

Thanks,
Florian

author maintainer
#14 lore raw

Re: [PATCH v10 1/4] random: add vgetrandom_alloc() syscall

Hi Florian,

Quoted 21 lines · Florian Weimer

* Jason A. Donenfeld:

Quoted 4 lines

I don't think zapping that memory is supported, or even a sensible thing to do. In the first place, I don't think we should suggest that the user can dereference that pointer, at all. In that sense, maybe it's best to call it a "handle" or something similar (a "HANDLE"! a "HWND"? a "HRNG"?

Surely the caller has to carve up the allocation, so the returned pointer is not opaque at all. From Adhemerval's glibc patch:

      grnd_allocator.cap = new_cap;
      grnd_allocator.states = new_states;

      for (size_t i = 0; i < num; ++i)
	{
	  grnd_allocator.states[i] = new_block;
	  new_block += size_per_each;
	}
      grnd_allocator.len = num;
    }
Quoted 1 line

That's the opposite of a handle, really.

Right. (And the same code is in the commit message example too.)

Quoted 12 lines
Quoted 12 lines
Quoted 7 lines

But it will constrain future
evolution of the implementation because you can't add registration (retaining a reference to the passed-in area in getrandom) after the fact. But I'm not sure if this is possible with the current interface, either. Userspace has to make some assumptions about the life-cycle to avoid a memory leak on thread exit.

It sounds like this is sort of a different angle on Rasmus' earlier comment about how munmap leaks implementation details. Maybe there's something to that after all? Or not? I see two approaches:

Quoted 7 lines
Quoted 4 lines

1) Keep munmap as the allocation function. If later on we do fancy registration and in-kernel state tracking, or add fancy protection flags, or whatever else, munmap should be able to identify these pages and carry out whatever special treatment is necessary.

munmap is fine, but the interface needs to say how to use it, and what length to pass.

Glad we're on the same page. Indeed I've now documented this for my in-progress v11. A blurb like:

+ * sys_vgetrandom_alloc - Allocate opaque states for use with vDSO getrandom().
+ *
+ * @num: On input, a pointer to a suggested hint of how many states to
+ * allocate, and on return the number of states actually allocated.
+ *
+ * @size_per_each: On input, must be zero. On return, the size of each state allocated,
+ * so that the caller can split up the returned allocation into
+ * individual states.
+ *
+ * @addr: Reserved, must be zero.
+ *
+ * @flags: Reserved, must be zero.
+ *
+ * The getrandom() vDSO function in userspace requires an opaque state, which
+ * this function allocates by mapping a certain number of special pages into
+ * the calling process. It takes a hint as to the number of opaque states
+ * desired, and provides the caller with the number of opaque states actually
+ * allocated, the size of each one in bytes, and the address of the first
+ * state, which may be split up into @num states of @size_per_each bytes each,
+ * by adding @size_per_each to the returned first state @num times.
+ *
+ * Returns the address of the first state in the allocation on success, or a
+ * negative error value on failure.
+ *
+ * The returned address of the first state may be passed to munmap(2) with a
+ * length of (size_t)num * (size_t)size_per_each, in order to deallocate the
+ * memory, after which it is invalid to pass it to vDSO getrandom().

What do you think of that text?

Quoted 7 lines
Quoted 5 lines

Then they're caught holding the bag? This doesn't seem much different from userspace shooting themselves in general, like writing garbage into the allocated states and then trying to use them. If this is something you really, really are concerned about, then maybe my cheesy dumb xor thing mentioned above would be a low effort mitigation here.

So the MAP_LOCKED is just there to prevent leakage to swap?

Right. I can combine that with MLOCK_ONFAULT and NORESERVED to avoid having to commit the memory immediately. I've got this in my tree for v11.

In case you're curious to see the WIP, it's in here:
https://git.zx2c4.com/linux-rng/log/?h=vdso

Jason

author maintainer
#15 lore raw

PATCH v10 2/4 arch: allocate vgetrandom_alloc() syscall number

In reply to Jason A. Donenfeld

Add vgetrandom_alloc() as syscall 451 (or 561 on alpha) by adding it to all of the various syscall.tbl and unistd.h files.

---
 arch/alpha/kernel/syscalls/syscall.tbl              | 1 +
 arch/arm/tools/syscall.tbl                          | 1 +
 arch/arm64/include/asm/unistd.h                     | 2 +-
 arch/arm64/include/asm/unistd32.h                   | 2 ++
 arch/ia64/kernel/syscalls/syscall.tbl               | 1 +
 arch/m68k/kernel/syscalls/syscall.tbl               | 1 +
 arch/microblaze/kernel/syscalls/syscall.tbl         | 1 +
 arch/mips/kernel/syscalls/syscall_n32.tbl           | 1 +
 arch/mips/kernel/syscalls/syscall_n64.tbl           | 1 +
 arch/mips/kernel/syscalls/syscall_o32.tbl           | 1 +
 arch/parisc/kernel/syscalls/syscall.tbl             | 1 +
 arch/powerpc/kernel/syscalls/syscall.tbl            | 1 +
 arch/s390/kernel/syscalls/syscall.tbl               | 1 +
 arch/sh/kernel/syscalls/syscall.tbl                 | 1 +
 arch/sparc/kernel/syscalls/syscall.tbl              | 1 +
 arch/x86/entry/syscalls/syscall_32.tbl              | 1 +
 arch/x86/entry/syscalls/syscall_64.tbl              | 1 +
 arch/xtensa/kernel/syscalls/syscall.tbl             | 1 +
 include/uapi/asm-generic/unistd.h                   | 5 ++++-
 tools/include/uapi/asm-generic/unistd.h             | 5 ++++-
 tools/perf/arch/mips/entry/syscalls/syscall_n64.tbl | 1 +
 tools/perf/arch/powerpc/entry/syscalls/syscall.tbl  | 1 +
 tools/perf/arch/s390/entry/syscalls/syscall.tbl     | 1 +
 tools/perf/arch/x86/entry/syscalls/syscall_64.tbl   | 1 +
 24 files changed, 31 insertions(+), 3 deletions(-)
arch/alpha/kernel/syscalls/syscall.tbl +1 -0
index 8ebacf37a8cf..a4bfd7b53d6f 100644
@@ -490,3 +490,4 @@
 558	common	process_mrelease		sys_process_mrelease
 559	common  futex_waitv                     sys_futex_waitv
 560	common	set_mempolicy_home_node		sys_ni_syscall
+561	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/arm/tools/syscall.tbl +1 -0
index ac964612d8b0..e10319cc6c3e 100644
@@ -464,3 +464,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common	futex_waitv			sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/arm64/include/asm/unistd.h +1 -1
index 037feba03a51..64a514f90131 100644
@@ -39,7 +39,7 @@
 #define __ARM_NR_compat_set_tls		(__ARM_NR_COMPAT_BASE + 5)
 #define __ARM_NR_COMPAT_END		(__ARM_NR_COMPAT_BASE + 0x800)
 
-#define __NR_compat_syscalls		451
+#define __NR_compat_syscalls		452
 #endif
 
 #define __ARCH_WANT_SYS_CLONE
arch/arm64/include/asm/unistd32.h +2 -0
index 604a2053d006..7285b5a830cc 100644
@@ -907,6 +907,8 @@ __SYSCALL(__NR_process_mrelease, sys_process_mrelease)
 __SYSCALL(__NR_futex_waitv, sys_futex_waitv)
 #define __NR_set_mempolicy_home_node 450
 __SYSCALL(__NR_set_mempolicy_home_node, sys_set_mempolicy_home_node)
+#define __NR_vgetrandom_alloc 451
+__SYSCALL(__NR_vgetrandom_alloc, sys_vgetrandom_alloc)
 
 /*
  * Please add new compat syscalls above this comment and update
arch/ia64/kernel/syscalls/syscall.tbl +1 -0
index 72c929d9902b..5ed9667051fc 100644
@@ -371,3 +371,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/m68k/kernel/syscalls/syscall.tbl +1 -0
index b1f3940bc298..d9e7ea26dd26 100644
@@ -450,3 +450,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/microblaze/kernel/syscalls/syscall.tbl +1 -0
index 820145e47350..c109e307a37b 100644
@@ -456,3 +456,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/mips/kernel/syscalls/syscall_n32.tbl +1 -0
index 253ff994ed2e..6d47d8231f7d 100644
@@ -389,3 +389,4 @@
 448	n32	process_mrelease		sys_process_mrelease
 449	n32	futex_waitv			sys_futex_waitv
 450	n32	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	n32	vgetrandom_alloc		sys_vgetrandom_alloc
arch/mips/kernel/syscalls/syscall_n64.tbl +1 -0
index 3f1886ad9d80..890e5b51e1fc 100644
@@ -365,3 +365,4 @@
 448	n64	process_mrelease		sys_process_mrelease
 449	n64	futex_waitv			sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	n64	vgetrandom_alloc		sys_vgetrandom_alloc
arch/mips/kernel/syscalls/syscall_o32.tbl +1 -0
index 8f243e35a7b2..de512de148f5 100644
@@ -438,3 +438,4 @@
 448	o32	process_mrelease		sys_process_mrelease
 449	o32	futex_waitv			sys_futex_waitv
 450	o32	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	o32	vgetrandom_alloc		sys_vgetrandom_alloc
arch/parisc/kernel/syscalls/syscall.tbl +1 -0
index 8a99c998da9b..bab1cee627e3 100644
@@ -448,3 +448,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common	futex_waitv			sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/powerpc/kernel/syscalls/syscall.tbl +1 -0
index a0be127475b1..e6c04eda2363 100644
@@ -537,3 +537,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450 	nospu	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/s390/kernel/syscalls/syscall.tbl +1 -0
index 799147658dee..5b0b2bea46da 100644
@@ -453,3 +453,4 @@
 448  common	process_mrelease	sys_process_mrelease		sys_process_mrelease
 449  common	futex_waitv		sys_futex_waitv			sys_futex_waitv
 450  common	set_mempolicy_home_node	sys_set_mempolicy_home_node	sys_set_mempolicy_home_node
+451  common	vgetrandom_alloc	sys_vgetrandom_alloc		sys_vgetrandom_alloc
arch/sh/kernel/syscalls/syscall.tbl +1 -0
index 2de85c977f54..631f0bac0e9a 100644
@@ -453,3 +453,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/sparc/kernel/syscalls/syscall.tbl +1 -0
index 4398cc6fb68d..b4925978adea 100644
@@ -496,3 +496,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
arch/x86/entry/syscalls/syscall_32.tbl +1 -0
index 320480a8db4f..f5f863a33824 100644
@@ -455,3 +455,4 @@
 448	i386	process_mrelease	sys_process_mrelease
 449	i386	futex_waitv		sys_futex_waitv
 450	i386	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	i386	vgetrandom_alloc		sys_vgetrandom_alloc
arch/x86/entry/syscalls/syscall_64.tbl +1 -0
index c84d12608cd2..0186f173f0e8 100644
@@ -372,6 +372,7 @@
 448	common	process_mrelease	sys_process_mrelease
 449	common	futex_waitv		sys_futex_waitv
 450	common	set_mempolicy_home_node	sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc	sys_vgetrandom_alloc
 
 #
 # Due to a historical design error, certain syscalls are numbered differently
arch/xtensa/kernel/syscalls/syscall.tbl +1 -0
index 52c94ab5c205..14d63a119cc2 100644
@@ -421,3 +421,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
include/uapi/asm-generic/unistd.h +4 -1
index 45fa180cc56a..9d2e299f3e8a 100644
@@ -886,8 +886,11 @@ __SYSCALL(__NR_futex_waitv, sys_futex_waitv)
 #define __NR_set_mempolicy_home_node 450
 __SYSCALL(__NR_set_mempolicy_home_node, sys_set_mempolicy_home_node)
 
+#define __NR_vgetrandom_alloc 451
+__SYSCALL(__NR_vgetrandom_alloc, sys_vgetrandom_alloc)
+
 #undef __NR_syscalls
-#define __NR_syscalls 451
+#define __NR_syscalls 452
 
 /*
  * 32 bit systems traditionally used different
tools/include/uapi/asm-generic/unistd.h +4 -1
index 45fa180cc56a..9d2e299f3e8a 100644
@@ -886,8 +886,11 @@ __SYSCALL(__NR_futex_waitv, sys_futex_waitv)
 #define __NR_set_mempolicy_home_node 450
 __SYSCALL(__NR_set_mempolicy_home_node, sys_set_mempolicy_home_node)
 
+#define __NR_vgetrandom_alloc 451
+__SYSCALL(__NR_vgetrandom_alloc, sys_vgetrandom_alloc)
+
 #undef __NR_syscalls
-#define __NR_syscalls 451
+#define __NR_syscalls 452
 
 /*
  * 32 bit systems traditionally used different
tools/perf/arch/mips/entry/syscalls/syscall_n64.tbl +1 -0
index 3f1886ad9d80..890e5b51e1fc 100644
@@ -365,3 +365,4 @@
 448	n64	process_mrelease		sys_process_mrelease
 449	n64	futex_waitv			sys_futex_waitv
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	n64	vgetrandom_alloc		sys_vgetrandom_alloc
tools/perf/arch/powerpc/entry/syscalls/syscall.tbl +1 -0
index e9e0df4f9a61..d58da67a9766 100644
@@ -534,3 +534,4 @@
 448	common	process_mrelease		sys_process_mrelease
 449	common  futex_waitv                     sys_futex_waitv
 450 	nospu	set_mempolicy_home_node		sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc		sys_vgetrandom_alloc
tools/perf/arch/s390/entry/syscalls/syscall.tbl +1 -0
index 799147658dee..5b0b2bea46da 100644
@@ -453,3 +453,4 @@
 448  common	process_mrelease	sys_process_mrelease		sys_process_mrelease
 449  common	futex_waitv		sys_futex_waitv			sys_futex_waitv
 450  common	set_mempolicy_home_node	sys_set_mempolicy_home_node	sys_set_mempolicy_home_node
+451  common	vgetrandom_alloc	sys_vgetrandom_alloc		sys_vgetrandom_alloc
tools/perf/arch/x86/entry/syscalls/syscall_64.tbl +1 -0
index c84d12608cd2..0186f173f0e8 100644
@@ -372,6 +372,7 @@
 448	common	process_mrelease	sys_process_mrelease
 449	common	futex_waitv		sys_futex_waitv
 450	common	set_mempolicy_home_node	sys_set_mempolicy_home_node
+451	common	vgetrandom_alloc	sys_vgetrandom_alloc
 
 #
 # Due to a historical design error, certain syscalls are numbered differently
Signature · 1 line
2.38.1
maintainer Acked-by on v10 2/4
#16 lore raw

Re: [PATCH v10 2/4] arch: allocate vgetrandom_alloc() syscall number

Hi Jason,

Quoted 4 lines · Jason A. Donenfeld

Add vgetrandom_alloc() as syscall 451 (or 561 on alpha) by adding it to all of the various syscall.tbl and unistd.h files.

Thanks for your patch!

What's the policy regarding adding syscall numbers for VDSO-related syscalls on architectures that do not support VDSOs yet?

Quoted 1 line
 arch/m68k/kernel/syscalls/syscall.tbl               | 1 +

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

author maintainer
#17 lore raw

Re: [PATCH v10 2/4] arch: allocate vgetrandom_alloc() syscall number

Hi Geert,

Quoted 6 lines · Geert Uytterhoeven

Hi Jason,

Quoted 3 lines · Jason A. Donenfeld

Add vgetrandom_alloc() as syscall 451 (or 561 on alpha) by adding it to all of the various syscall.tbl and unistd.h files.

Quoted 6 lines
Quoted 1 line

Thanks for your patch!

What's the policy regarding adding syscall numbers for VDSO-related syscalls on architectures that do not support VDSOs yet?

I don't know exactly what the /policy/ is, but not wanting to rock the boat, the first iterations of this series only added it to x86. But then Arnd joined the thread and said I should add it to all of them all at once and separate that out into this commit, so that's what we have here. I think the idea is to keep syscall numbers synchronized these days between archs if possible.

Jason

maintainer
#18 lore raw

Re: [PATCH v10 2/4] arch: allocate vgetrandom_alloc() syscall number

Quoted 14 lines · Jason A. Donenfeld
Quoted 6 lines · Geert Uytterhoeven

Hi Jason,

Quoted 3 lines · Jason A. Donenfeld

Add vgetrandom_alloc() as syscall 451 (or 561 on alpha) by adding it to all of the various syscall.tbl and unistd.h files.

Quoted 6 lines
Quoted 1 line

Thanks for your patch!

What's the policy regarding adding syscall numbers for VDSO-related syscalls on architectures that do not support VDSOs yet?

Quoted 6 lines

I don't know exactly what the /policy/ is, but not wanting to rock the boat, the first iterations of this series only added it to x86. But then Arnd joined the thread and said I should add it to all of them all at once and separate that out into this commit, so that's what we have here. I think the idea is to keep syscall numbers synchronized these days between archs if possible.

Right, it shouldn't matter if the syscall has anything to do with vdso or some other feature, the important bit is that this is an optional syscall that may or may not exist in a kernel.

Adding every new syscall number to all architectures helps avoid merge conflicts and keeps the numbers synchronized. It's entirely plausible that this one gets enabled on an architecture that starts out with the default -ENOSYS implementation, and has that backported to 6.2 (or even older kernels in principle) for a distro release, so it also makes sense to have it in the uapi table as soon as we have fixed the number.

Arnd

author maintainer
#19 lore raw

PATCH v10 3/4 random: introduce generic vDSO getrandom() implementation

In reply to Jason A. Donenfeld

Provide a generic C vDSO getrandom() implementation, which operates on an opaque state returned by vgetrandom_alloc() and produces random bytes the same way as getrandom(). This has a the API signature:

  ssize_t vgetrandom(void *buffer, size_t len, unsigned int flags, void *opaque_state);

The return value and the first 3 arguments are the same as ordinary getrandom(), while the last argument is a pointer to the opaque allocated state. Were all four arguments passed to the getrandom() syscall, nothing different would happen, and the functions would have the exact same behavior.

The actual vDSO RNG algorithm implemented is the same one implemented by drivers/char/random.c, using the same fast-erasure techniques as that. Should the in-kernel implementation change, so too will the vDSO one.

It requires an implementation of ChaCha20 that does not use any stack, in order to maintain forward secrecy if a multi-threaded program forks (though this does not account for a similar issue with SA_SIGINFO copying registers to the stack), so this is left as an architecture-specific fill-in. Stack-less ChaCha20 is an easy algorithm to implement on a variety of architectures, so this shouldn't be too onerous.

Initially, the state is keyless, and so the first call makes a getrandom() syscall to generate that key, and then uses it for subsequent calls. By keeping track of a generation counter, it knows when its key is invalidated and it should fetch a new one using the syscall. Later, more than just a generation counter might be used.

Since MADV_WIPEONFORK is set on the opaque state, the key and related state is wiped during a fork(), so secrets don't roll over into new processes, and the same state doesn't accidentally generate the same random stream. The generation counter, as well, is always >0, so that the 0 counter is a useful indication of a fork() or otherwise uninitialized state.

If the kernel RNG is not yet initialized, then the vDSO always calls the syscall, because that behavior cannot be emulated in userspace, but fortunately that state is short lived and only during early boot. If it has been initialized, then there is no need to inspect the flags argument, because the behavior does not change post-initialization regardless of the flags value.

Since the opaque state passed to it is mutated, vDSO getrandom() is not reentrant, when used with the same opaque state, which libc should be mindful of.

vgetrandom_alloc() and vDSO getrandom() together provide the ability for userspace to generate random bytes quickly and safely, and is intended to be integrated into libc's thread management. As an illustrative example, the following code might be used to do the same outside of libc. All of the static functions are to be considered implementation private, including the vgetrandom_alloc() syscall wrapper, which generally shouldn't be exposed outside of libc, with the non-static vgetrandom() function at the end being the exported interface. The various pthread-isms are expected to be elided into libc internals. This per-thread allocation scheme is very naive and does not shrink; other implementations may choose to be more complex.

  static void *vgetrandom_alloc(unsigned int *num, unsigned int *size_per_each, unsigned int flags)
  {
    long ret = syscall(__NR_vgetrandom_alloc, &num, &size_per_each, flags);
    return ret == -1 ? NULL : (void *)ret;
  }

  static struct {
    pthread_mutex_t lock;
    void **states;
    size_t len, cap;
  } grnd_allocator = {
    .lock = PTHREAD_MUTEX_INITIALIZER
  };

  static void *vgetrandom_get_state(void)
  {
    void *state = NULL;
    pthread_mutex_lock(&grnd_allocator.lock);
    if (!grnd_allocator.len) {
      size_t new_cap;
      unsigned int size_per_each, num = 16; /* Just a hint. Could also be nr_cpus. */
      void *new_block = vgetrandom_alloc(&num, &size_per_each, 0), *new_states;

      if (!new_block)
        goto out;
      new_cap = grnd_allocator.cap + num;
      new_states = reallocarray(grnd_allocator.states, new_cap, sizeof(*grnd_allocator.states));
      if (!new_states) {
        munmap(new_block, num * size_per_each);
        goto out;
      }
      grnd_allocator.cap = new_cap;
      grnd_allocator.states = new_states;

      for (size_t i = 0; i < num; ++i) {
        grnd_allocator.states[i] = new_block;
        new_block += size_per_each;
      }
      grnd_allocator.len = num;
    }
    state = grnd_allocator.states[--grnd_allocator.len];
  out:
    pthread_mutex_unlock(&grnd_allocator.lock);
    return state;
  }

  static void vgetrandom_put_state(void *state)
  {
    if (!state)
      return;
    pthread_mutex_lock(&grnd_allocator.lock);
    grnd_allocator.states[grnd_allocator.len++] = state;
    pthread_mutex_unlock(&grnd_allocator.lock);
  }

  static struct {
    ssize_t(*fn)(void *buf, size_t len, unsigned long flags, void *state);
    pthread_key_t key;
    pthread_once_t initialized;
  } grnd_ctx = {
    .initialized = PTHREAD_ONCE_INIT
  };

  static void vgetrandom_init(void)
  {
    if (pthread_key_create(&grnd_ctx.key, vgetrandom_put_state) != 0)
      return;
    grnd_ctx.fn = __vdsosym("LINUX_2.6", "__vdso_getrandom");
  }

  ssize_t vgetrandom(void *buf, size_t len, unsigned long flags)
  {
    void *state;
    pthread_once(&grnd_ctx.initialized, vgetrandom_init);
    if (!grnd_ctx.fn)
      return getrandom(buf, len, flags);
    state = pthread_getspecific(grnd_ctx.key);
    if (!state) {
      state = vgetrandom_get_state();
      if (pthread_setspecific(grnd_ctx.key, state) != 0) {
        vgetrandom_put_state(state);
        state = NULL;
      }
      if (!state)
        return getrandom(buf, len, flags);
    }
    return grnd_ctx.fn(buf, len, flags, state);

}

---
 MAINTAINERS             |   1 +
 drivers/char/random.c   |   9 ++
 include/vdso/datapage.h |  11 +++
 lib/vdso/Kconfig        |   7 +-
 lib/vdso/getrandom.c    | 204 ++++++++++++++++++++++++++++++++++++++++
 5 files changed, 231 insertions(+), 1 deletion(-)
 create mode 100644 lib/vdso/getrandom.c
MAINTAINERS +1 -0
index 3894f947a507..70dff39fcff9 100644
@@ -17288,6 +17288,7 @@ S:	Maintained
 F:	drivers/char/random.c
 F:	drivers/virt/vmgenid.c
 F:	include/vdso/getrandom.h
+F:	lib/vdso/getrandom.c
 
 RAPIDIO SUBSYSTEM
 M:	Matt Porter <[email protected]>
drivers/char/random.c +9 -0
index b81d67f3ebab..b37a3f9367a9 100644
@@ -60,6 +60,9 @@
 #ifdef CONFIG_VGETRANDOM_ALLOC_SYSCALL
 #include <vdso/getrandom.h>
 #endif
+#ifdef CONFIG_VDSO_GETRANDOM
+#include <vdso/datapage.h>
+#endif
 #include <asm/processor.h>
 #include <asm/irq.h>
 #include <asm/irq_regs.h>
@@ -344,6 +347,9 @@ static void crng_reseed(struct work_struct *work)
 	if (next_gen == ULONG_MAX)
 		++next_gen;
 	WRITE_ONCE(base_crng.generation, next_gen);
+#ifdef CONFIG_VDSO_GETRANDOM
+	smp_store_release(&_vdso_rng_data.generation, next_gen + 1);
+#endif
 	if (!static_branch_likely(&crng_is_ready))
 		crng_init = CRNG_READY;
 	spin_unlock_irqrestore(&base_crng.lock, flags);
@@ -794,6 +800,9 @@ static void __cold _credit_init_bits(size_t bits)
 		if (static_key_initialized)
 			execute_in_process_context(crng_set_ready, &set_ready);
 		atomic_notifier_call_chain(&random_ready_notifier, 0, NULL);
+#ifdef CONFIG_VDSO_GETRANDOM
+		smp_store_release(&_vdso_rng_data.is_ready, true);
+#endif
 		wake_up_interruptible(&crng_init_wait);
 		kill_fasync(&fasync, SIGIO, POLL_IN);
 		pr_notice("crng init done\n");
include/vdso/datapage.h +11 -0
index 73eb622e7663..9ae4d76b36c7 100644
@@ -109,6 +109,16 @@ struct vdso_data {
 	struct arch_vdso_data	arch_data;
 };
 
+/**
+ * struct vdso_rng_data - vdso RNG state information
+ * @generation:	a counter representing the number of RNG reseeds
+ * @is_ready:	whether the RNG is initialized
+ */
+struct vdso_rng_data {
+	unsigned long	generation;
+	bool		is_ready;
+};
+
 /*
  * We use the hidden visibility to prevent the compiler from generating a GOT
  * relocation. Not only is going through a GOT useless (the entry couldn't and
@@ -120,6 +130,7 @@ struct vdso_data {
  */
 extern struct vdso_data _vdso_data[CS_BASES] __attribute__((visibility("hidden")));
 extern struct vdso_data _timens_data[CS_BASES] __attribute__((visibility("hidden")));
+extern struct vdso_rng_data _vdso_rng_data __attribute__((visibility("hidden")));
 
 /*
  * The generic vDSO implementation requires that gettimeofday.h
lib/vdso/Kconfig +6 -1
index b22584f8da03..f12b76642921 100644
@@ -29,7 +29,6 @@ config GENERIC_VDSO_TIME_NS
 	help
 	  Selected by architectures which support time namespaces in the
 	  VDSO
-
 endif
 
 config VGETRANDOM_ALLOC_SYSCALL
@@ -38,3 +37,9 @@ config VGETRANDOM_ALLOC_SYSCALL
 	help
 	  Selected by the getrandom() vDSO function, which requires this
 	  for state allocation.
+
+config VDSO_GETRANDOM
+	bool
+	select VGETRANDOM_ALLOC_SYSCALL
+	help
+	  Selected by architectures that support vDSO getrandom().
lib/vdso/getrandom.c new file +204 -0

204 lines not on this page · raw message

Signature · 1 line
2.38.1
#20 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Jason!

Quoted 3 lines · Jason A. Donenfeld

+/**
+ * struct vdso_rng_data - vdso RNG state information
+ * @generation: a counter representing the number of RNG reseeds

A counter

Quoted 1 line

+ * @is_ready: whether the RNG is initialized

Signals whether ...

Quoted 6 lines

+ */

(quoted diff) +5 -0
+struct vdso_rng_data {
+	unsigned long	generation;
+	bool		is_ready;
+};
+

Review

Quoted 10 lines
(quoted diff) +10 -0
+
+#define MEMCPY_AND_ZERO_SRC(type, dst, src, len) do { \
+	while (len >= sizeof(type)) { \
+		__put_unaligned_t(type, __get_unaligned_t(type, src), dst); \
+		__put_unaligned_t(type, 0, src); \
+		dst += sizeof(type); \
+		src += sizeof(type); \
+		len -= sizeof(type); \
+	} \
+} while (0)

I'd appreciate it if you go back to the code I suggested to you and compare and contrast it in terms of readability.

Quoted 9 lines
lib/vdso/getrandom.c new file +9 -0
+
+static void memcpy_and_zero_src(void *dst, void *src, size_t len)
+{
+	if (IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS)) {
+		if (IS_ENABLED(CONFIG_64BIT))
+			MEMCPY_AND_ZERO_SRC(u64, dst, src, len);
+		MEMCPY_AND_ZERO_SRC(u32, dst, src, len);
+		MEMCPY_AND_ZERO_SRC(u16, dst, src, len);
+	}

lib/vdso/getrandom.c line 41

Quoted 10 lines
lib/vdso/getrandom.c new file +10 -0
+	MEMCPY_AND_ZERO_SRC(u8, dst, src, len);
+}
+
+/**
+ * __cvdso_getrandom_data - generic vDSO implementation of getrandom() syscall
+ * @rng_info:		describes state of kernel RNG, memory shared with kernel
+ * @buffer:		destination buffer to fill with random bytes
+ * @len:		size of @buffer in bytes
+ * @flags:		zero or more GRND_* flags
+ * @opaque_state:	a pointer to an opaque state area

NIT. Please start the explanations with an uppercase letter

lib/vdso/getrandom.c line 142

Quoted 7 lines
lib/vdso/getrandom.c new file +7 -0
+		/*
+		 * Set @state->pos to beyond the end of the batch, so that the batch is refilled
+		 * using the new key.
+		 */
+		state->pos = sizeof(state->batch);
+	}
+

This one is odd:

lib/vdso/getrandom.c line 143

Quoted 1 line
lib/vdso/getrandom.c new file +1 -0
+	len = ret;

@ret is not modified after the initialization at the top of the function:

lib/vdso/getrandom.c line 59

Quoted 1 line
lib/vdso/getrandom.c new file +1 -0
+	ssize_t ret = min_t(size_t, INT_MAX & PAGE_MASK /* = MAX_RW_COUNT */, len);

so I really had to go up a page and figure out what the story is.

lib/vdso/getrandom.c line 194

Quoted 4 lines
lib/vdso/getrandom.c new file +4 -0
+
+	/* Since the batch was just refilled, set the position back to 0 to indicate a full batch. */
+	state->pos = 0;
+	goto more_batch;

Aside of the nitpicks above, thank you very much for making this comprehensible.

The comments are well done and appreciated and I'm pretty sure that this part:

lib/vdso/getrandom.c line 94

Quoted 4 lines
lib/vdso/getrandom.c new file +4 -0
+	in_use = READ_ONCE(state->in_use);
+	if (unlikely(in_use))
+		goto fallback_syscall;
+	WRITE_ONCE(state->in_use, true);

was very much induced by writing those comments :)

Thanks,

tglx

author maintainer
#21 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Hi Thomas,

Quoted 6 lines · Thomas Gleixner

Jason!

Quoted 3 lines · Jason A. Donenfeld

+/**
+ * struct vdso_rng_data - vdso RNG state information
+ * @generation: a counter representing the number of RNG reseeds

FYI, ever other struct in this file uses lower case and no punctuation, so I'll follow that for this one.

Quoted 5 lines

A counter

Quoted 1 line

+ * @is_ready: whether the RNG is initialized

Signals whether ...

Ack.

Quoted 11 lines
Quoted 11 lines

+ */

include/vdso/datapage.h +10 -0
+struct vdso_rng_data {
+	unsigned long	generation;
+	bool		is_ready;
+};
+
+
+#define MEMCPY_AND_ZERO_SRC(type, dst, src, len) do { \
+	while (len >= sizeof(type)) { \
+		__put_unaligned_t(type, __get_unaligned_t(type, src), dst); \
+		__put_unaligned_t(type, 0, src); \
Quoted 8 lines

lib/vdso/getrandom.c line 22

Quoted 5 lines
lib/vdso/getrandom.c new file +5 -0
+		dst += sizeof(type); \
+		src += sizeof(type); \
+		len -= sizeof(type); \
+	} \
+} while (0)

I'd appreciate it if you go back to the code I suggested to you and compare and contrast it in terms of readability.

Ahh, you like to align your \. Okay, I'll do that. I also added a do { ... } while (0) wrapper around it, but I think it makes sense to keep that so that there aren't stray semicolons.

Quoted 9 lines
Quoted 7 lines

+/**
+ * __cvdso_getrandom_data - generic vDSO implementation of getrandom() syscall
+ * @rng_info: describes state of kernel RNG, memory shared with kernel
+ * @buffer: destination buffer to fill with random bytes
+ * @len: size of @buffer in bytes
+ * @flags: zero or more GRND_* flags
+ * @opaque_state: a pointer to an opaque state area

NIT. Please start the explanations with an uppercase letter

Okay. Will do everywhere in this patchset except for in vdso/datapage.h.

Quoted 10 lines

This one is odd:

lib/vdso/getrandom.c line 143

Quoted 1 line
lib/vdso/getrandom.c new file +1 -0
+	len = ret;

@ret is not modified after the initialization at the top of the function:

lib/vdso/getrandom.c line 59

Quoted 1 line
lib/vdso/getrandom.c new file +1 -0
+	ssize_t ret = min_t(size_t, INT_MAX & PAGE_MASK /* = MAX_RW_COUNT */, len);

so I really had to go up a page and figure out what the story is.

If the generation changes, and it's tried again, the whole random buffer is filled again, so that has to be reset. I'll leave a comment.

Quoted 7 lines

lib/vdso/getrandom.c line 194

Quoted 4 lines
lib/vdso/getrandom.c new file +4 -0
+
+	/* Since the batch was just refilled, set the position back to 0 to indicate a full batch. */
+	state->pos = 0;
+	goto more_batch;

Aside of the nitpicks above, thank you very much for making this comprehensible.

Thanks for nudging me in the right direction.

Quoted 10 lines

The comments are well done and appreciated and I'm pretty sure that this part:

lib/vdso/getrandom.c line 94

Quoted 4 lines
lib/vdso/getrandom.c new file +4 -0
+	in_use = READ_ONCE(state->in_use);
+	if (unlikely(in_use))
+		goto fallback_syscall;
+	WRITE_ONCE(state->in_use, true);

was very much induced by writing those comments :)

Well, not exactly, unfortunately. Adhemerval -- the glibc maintainer working on the libc side of this -- and I have been discussing signal handling craziness and lots of different schemes over the last week+, and this rather simple thing is the result of those conversations.

Jason

#22 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

In reply to Jason A. Donenfeld

* Jason A. Donenfeld:

Quoted 8 lines
include/vdso/datapage.h +0 -0
index 73eb622e7663..9ae4d76b36c7 100644
@@ -109,6 +109,16 @@ struct vdso_data {
 	struct arch_vdso_data	arch_data;
 };
 

include/vdso/datapage.h struct vdso_data { · line 121

Quoted 10 lines
include/vdso/datapage.h +10 -0
+/**
+ * struct vdso_rng_data - vdso RNG state information
+ * @generation:	a counter representing the number of RNG reseeds
+ * @is_ready:	whether the RNG is initialized
+ */
+struct vdso_rng_data {
+	unsigned long	generation;
+	bool		is_ready;
+};
+

I don't think you can use a type like long here. The header says this:

* vdso_data will be accessed by 64 bit and compat code at the same time
* so we should be careful before modifying this structure.

So the ABI must be same for 32-bit and 64-bit mode, and long isn't.

Thanks,
Florian

author maintainer
#23 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Hi Florian,

Quoted 23 lines · Florian Weimer

* Jason A. Donenfeld:

Quoted 8 lines
include/vdso/datapage.h +0 -0
index 73eb622e7663..9ae4d76b36c7 100644
@@ -109,6 +109,16 @@ struct vdso_data {
 	struct arch_vdso_data	arch_data;
 };
 

include/vdso/datapage.h struct vdso_data { · line 121

Quoted 10 lines
include/vdso/datapage.h +10 -0
+/**
+ * struct vdso_rng_data - vdso RNG state information
+ * @generation:	a counter representing the number of RNG reseeds
+ * @is_ready:	whether the RNG is initialized
+ */
+struct vdso_rng_data {
+	unsigned long	generation;
+	bool		is_ready;
+};
+

I don't think you can use a type like long here. The header says this:

Quoted 4 lines

* vdso_data will be accessed by 64 bit and compat code at the same time
* so we should be careful before modifying this structure.

So the ABI must be same for 32-bit and 64-bit mode, and long isn't.

Excellent point. The size of the type needs to be explicit. Will fix.

Jason

author maintainer
#24 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Quoted 26 lines · Jason A. Donenfeld

Hi Florian,

Quoted 23 lines · Florian Weimer

* Jason A. Donenfeld:

Quoted 8 lines
include/vdso/datapage.h +0 -0
index 73eb622e7663..9ae4d76b36c7 100644
@@ -109,6 +109,16 @@ struct vdso_data {
 	struct arch_vdso_data	arch_data;
 };
 

include/vdso/datapage.h struct vdso_data { · line 121

Quoted 10 lines
include/vdso/datapage.h +10 -0
+/**
+ * struct vdso_rng_data - vdso RNG state information
+ * @generation:	a counter representing the number of RNG reseeds
+ * @is_ready:	whether the RNG is initialized
+ */
+struct vdso_rng_data {
+	unsigned long	generation;
+	bool		is_ready;
+};
+

I don't think you can use a type like long here. The header says this:

Quoted 6 lines
Quoted 4 lines

* vdso_data will be accessed by 64 bit and compat code at the same time
* so we should be careful before modifying this structure.

So the ABI must be same for 32-bit and 64-bit mode, and long isn't.

Excellent point. The size of the type needs to be explicit. Will fix.

I'll do something like this:

include/vdso/datapage.h +15 -2
index 80e25cdcdb1c..218bbeac5613 100644
@@ -19,6 +19,19 @@
 #include <vdso/time32.h>
 #include <vdso/time64.h>
 
+/**
+ * type vdso_kernel_ulong - unsigned long type that matches kernel's unsigned long
+ *
+ * The structs in this file must operate the same way in both 64-bit code and
+ * in 32-bit compat code, over the same potentially 64-bit kernel. So, this
+ * type represents the size of an unsigned long as used by kernel-space code.
+ */
+#ifdef CONFIG_64BIT
+typedef u64 vdso_kernel_ulong;
+#else
+typedef u32 vdso_kernel_ulong;
+#endif
+
 #ifdef CONFIG_ARCH_HAS_VDSO_DATA
 #include <asm/vdso/data.h>
 #else
@@ -115,8 +128,8 @@ struct vdso_data {
  * @is_ready:	signals whether the RNG is initialized
  */
 struct vdso_rng_data {
-	unsigned long	generation;
-	bool		is_ready;
+	vdso_kernel_ulong	generation;
+	bool			is_ready;
 };
 
 /*
maintainer
#25 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Quoted 11 lines · Jason A. Donenfeld
Quoted 6 lines · Jason A. Donenfeld
Quoted 5 lines · Florian Weimer

* vdso_data will be accessed by 64 bit and compat code at the same time
* so we should be careful before modifying this structure.

So the ABI must be same for 32-bit and 64-bit mode, and long isn't.

I'll do something like this:

include/vdso/datapage.h line 33

Quoted 5 lines
include/vdso/datapage.h +5 -0
+#ifdef CONFIG_64BIT
+typedef u64 vdso_kernel_ulong;
+#else
+typedef u32 vdso_kernel_ulong;
+#endif

This does not address the ABI concern: to allow 32-bit and 64-bit tasks to share the same data page, it has to be the same width on both, either u32 or 64, but not depending on a configuration option.

Quoted 4 lines
struct vdso_rng_data {
	vdso_kernel_ulong	generation;
	bool			is_ready;
};

There is another problem with this: you have implicit padding in the structure because the two members have different size and alignment requirements. The easiest fix is to make them both u64, or you could have a u32 is_ready and an explit u32 for the padding.

Arnd

author maintainer
#26 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Hi Arnd,

Quoted 10 lines · Arnd Bergmann

include/vdso/datapage.h line 33

Quoted 5 lines
include/vdso/datapage.h +5 -0
+#ifdef CONFIG_64BIT
+typedef u64 vdso_kernel_ulong;
+#else
+typedef u32 vdso_kernel_ulong;
+#endif

This does not address the ABI concern: to allow 32-bit and 64-bit tasks to share the same data page, it has to be the same width on both, either u32 or 64, but not depending on a configuration option.

I think it does address the issue. CONFIG_64BIT is a .config setting, not a compiler-derived setting. So a 64-bit kernel will get a u64 in kernel mode, and then it will get a u64 for the 64-bit vdso usermode compile, and finally it will get a u64 for the 32-bit vdso usermode compile. So in all three cases, the size is the same.

Quoted 10 lines
Quoted 4 lines
struct vdso_rng_data {
      vdso_kernel_ulong       generation;
      bool                    is_ready;
};

There is another problem with this: you have implicit padding in the structure because the two members have different size and alignment requirements. The easiest fix is to make them both u64, or you could have a u32 is_ready and an explit u32 for the padding.

There's padding at the end of the structure, yes. But both generation and is_ready will be at the same offset. If the structure grows, then sure, that'll have to be taken into account. But that's not a problem because this is a private implementation detail between the vdso code and the kernel.

Jason

maintainer
#27 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Quoted 14 lines · Jason A. Donenfeld

Hi Arnd,

Quoted 10 lines · Arnd Bergmann

include/vdso/datapage.h line 33

Quoted 5 lines
include/vdso/datapage.h +5 -0
+#ifdef CONFIG_64BIT
+typedef u64 vdso_kernel_ulong;
+#else
+typedef u32 vdso_kernel_ulong;
+#endif

This does not address the ABI concern: to allow 32-bit and 64-bit tasks to share the same data page, it has to be the same width on both, either u32 or 64, but not depending on a configuration option.

Quoted 5 lines

I think it does address the issue. CONFIG_64BIT is a .config setting, not a compiler-derived setting. So a 64-bit kernel will get a u64 in kernel mode, and then it will get a u64 for the 64-bit vdso usermode compile, and finally it will get a u64 for the 32-bit vdso usermode compile. So in all three cases, the size is the same.

I see what you mean now. However this means your vdso32 copies are different between 32-bit and 64-bit kernels. If you need to access one of the fields from assembler, it even ends up different at source level, which adds a bit of complexity.

Making the interface configuration-independent makes it obvious to the reader that none of these problems can happen.

Quoted 11 lines
Quoted 10 lines
Quoted 4 lines
struct vdso_rng_data {
      vdso_kernel_ulong       generation;
      bool                    is_ready;
};

There is another problem with this: you have implicit padding in the structure because the two members have different size and alignment requirements. The easiest fix is to make them both u64, or you could have a u32 is_ready and an explit u32 for the padding.

Quoted 5 lines

There's padding at the end of the structure, yes. But both generation and is_ready will be at the same offset. If the structure grows, then sure, that'll have to be taken into account. But that's not a problem because this is a private implementation detail between the vdso code and the kernel.

I was not concerned about incompatibility here, but rather about possibly leaking kernel data to the vdso page. Again, this probably doesn't happen if your code is written correctly, but the rule for kernel-user ABIs is to avoid implicit padding to ensure that the padding bytes can never leak any information. Using structures without padding at the minimum helps avoid having to think about whether this can become a problem when inspecting the code for possible issues, both from humans and from automated tools.

Arnd

author maintainer
#28 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Hi Arnd,

Quoted 6 lines · Arnd Bergmann
Quoted 5 lines

I think it does address the issue. CONFIG_64BIT is a .config setting, not a compiler-derived setting. So a 64-bit kernel will get a u64 in kernel mode, and then it will get a u64 for the 64-bit vdso usermode compile, and finally it will get a u64 for the 32-bit vdso usermode compile. So in all three cases, the size is the same.

Quoted 7 lines

I see what you mean now. However this means your vdso32 copies are different between 32-bit and 64-bit kernels. If you need to access one of the fields from assembler, it even ends up different at source level, which adds a bit of complexity.

Making the interface configuration-independent makes it obvious to the reader that none of these problems can happen.

Except ideally, these are word-sized accesses (where only compat code has to suffer I suppose).

Quoted 11 lines
Quoted 11 lines
Quoted 10 lines
Quoted 4 lines
struct vdso_rng_data {
      vdso_kernel_ulong       generation;
      bool                    is_ready;
};

There is another problem with this: you have implicit padding in the structure because the two members have different size and alignment requirements. The easiest fix is to make them both u64, or you could have a u32 is_ready and an explit u32 for the padding.

Quoted 8 lines
Quoted 5 lines

There's padding at the end of the structure, yes. But both generation and is_ready will be at the same offset. If the structure grows, then sure, that'll have to be taken into account. But that's not a problem because this is a private implementation detail between the vdso code and the kernel.

I was not concerned about incompatibility here, but rather about possibly leaking kernel data to the vdso page.

The vvar page starts out zeroed, no?

Jason

maintainer
#29 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Quoted 10 lines · Jason A. Donenfeld
Quoted 8 lines
Quoted 5 lines

There's padding at the end of the structure, yes. But both generation and is_ready will be at the same offset. If the structure grows, then sure, that'll have to be taken into account. But that's not a problem because this is a private implementation detail between the vdso code and the kernel.

I was not concerned about incompatibility here, but rather about possibly leaking kernel data to the vdso page.

The vvar page starts out zeroed, no?

The typical problem is someone doing a copy_to_user() of an in-kernel structure into the userspace side, which would then copy the padding as well. If the source is on the stack, a malicious caller can trick the another syscall into leaving sensitive data at this exact stack location. Again, I'm not saying that your code is vulnerable to that type of attack, just that making all ABI structures not have holes is useful for auditing.

Arnd

author maintainer
#30 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

Quoted 11 lines · Arnd Bergmann
Quoted 9 lines · Jason A. Donenfeld
Quoted 8 lines
Quoted 5 lines

There's padding at the end of the structure, yes. But both generation and is_ready will be at the same offset. If the structure grows, then sure, that'll have to be taken into account. But that's not a problem because this is a private implementation detail between the vdso code and the kernel.

I was not concerned about incompatibility here, but rather about possibly leaking kernel data to the vdso page.

Quoted 7 lines
Quoted 1 line

The vvar page starts out zeroed, no?

The typical problem is someone doing a copy_to_user() of an in-kernel structure into the userspace side, which would then copy the padding as well. If the source is on the stack, a malicious caller can trick the another syscall into leaving sensitive data at this exact stack location.

I'm quite aware of this infoleak, having made use of it countless times over the years. It just doesn't seem relevant to the vvar page.

Jason

#31 lore raw

Re: [PATCH v10 3/4] random: introduce generic vDSO getrandom() implementation

In reply to Jason A. Donenfeld
Quoted 11 lines · Jason A. Donenfeld
Quoted 7 lines · Arnd Bergmann

I see what you mean now. However this means your vdso32 copies are different between 32-bit and 64-bit kernels. If you need to access one of the fields from assembler, it even ends up different at source level, which adds a bit of complexity.

Making the interface configuration-independent makes it obvious to the reader that none of these problems can happen.

Except ideally, these are word-sized accesses (where only compat code has to suffer I suppose).

While I hate it with a passion, there is actually a valid reason to use this ugly typedef.

On 32bit architectures which have load/store tearing of 64bit variables into two 32bit accesses due to ISA limitations, that results in undefined behaviour when write and read are concurrent. Neither READ_ONCE() nor WRITE_ONCE help there.

Though that begs the question whether we need a 64bit generation counter for the VDSO at all.

Thanks,

tglx

author maintainer
#32 lore raw

PATCH v10 4/4 x86: vdso: Wire up getrandom() vDSO implementation

In reply to Jason A. Donenfeld

Hook up the generic vDSO implementation to the x86 vDSO data page. Since the existing vDSO infrastructure is heavily based on the timekeeping functionality, which works over arrays of bases, a new macro is introduced for vvars that are not arrays.

The vDSO function requires a ChaCha20 implementation that does not write to the stack, yet can still do an entire ChaCha20 permutation, so provide this using SSE2, since this is userland code that must work on all x86-64 processors.

---
 arch/x86/Kconfig                        |   1 +
 arch/x86/entry/vdso/Makefile            |   3 +-
 arch/x86/entry/vdso/vdso.lds.S          |   2 +
 arch/x86/entry/vdso/vgetrandom-chacha.S | 177 ++++++++++++++++++++++++
 arch/x86/entry/vdso/vgetrandom.c        |  17 +++
 arch/x86/include/asm/vdso/getrandom.h   |  55 ++++++++
 arch/x86/include/asm/vdso/vsyscall.h    |   2 +
 arch/x86/include/asm/vvar.h             |  16 +++
 8 files changed, 272 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/entry/vdso/vgetrandom-chacha.S
 create mode 100644 arch/x86/entry/vdso/vgetrandom.c
 create mode 100644 arch/x86/include/asm/vdso/getrandom.h
arch/x86/Kconfig +1 -0
index 67745ceab0db..357148c4a3a4 100644
@@ -269,6 +269,7 @@ config X86
 	select HAVE_UNSTABLE_SCHED_CLOCK
 	select HAVE_USER_RETURN_NOTIFIER
 	select HAVE_GENERIC_VDSO
+	select VDSO_GETRANDOM			if X86_64
 	select HOTPLUG_SMT			if SMP
 	select IRQ_FORCED_THREADING
 	select NEED_PER_CPU_EMBED_FIRST_CHUNK
arch/x86/entry/vdso/Makefile +2 -1
index 3e88b9df8c8f..2de64e52236a 100644
@@ -27,7 +27,7 @@ VDSO32-$(CONFIG_X86_32)		:= y
 VDSO32-$(CONFIG_IA32_EMULATION)	:= y
 
 # files to link into the vdso
-vobjs-y := vdso-note.o vclock_gettime.o vgetcpu.o
+vobjs-y := vdso-note.o vclock_gettime.o vgetcpu.o vgetrandom.o vgetrandom-chacha.o
 vobjs32-y := vdso32/note.o vdso32/system_call.o vdso32/sigreturn.o
 vobjs32-y += vdso32/vclock_gettime.o
 vobjs-$(CONFIG_X86_SGX)	+= vsgx.o
@@ -104,6 +104,7 @@ CFLAGS_REMOVE_vclock_gettime.o = -pg
 CFLAGS_REMOVE_vdso32/vclock_gettime.o = -pg
 CFLAGS_REMOVE_vgetcpu.o = -pg
 CFLAGS_REMOVE_vsgx.o = -pg
+CFLAGS_REMOVE_vgetrandom.o = -pg
 
 #
 # X32 processes use x32 vDSO to access 64bit kernel data.
arch/x86/entry/vdso/vdso.lds.S +2 -0
index 4bf48462fca7..1919cc39277e 100644
@@ -28,6 +28,8 @@ VERSION {
 		clock_getres;
 		__vdso_clock_getres;
 		__vdso_sgx_enter_enclave;
+		getrandom;
+		__vdso_getrandom;
 	local: *;
 	};
 }
arch/x86/entry/vdso/vgetrandom-chacha.S new file +177 -0

177 lines not on this page · raw message

arch/x86/entry/vdso/vgetrandom.c new file +17 -0
new file mode 100644
index 000000000000..6045ded5da90
@@ -0,0 +1,17 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2022 Jason A. Donenfeld <[email protected]>. All Rights Reserved.
+ */
+#include <linux/types.h>
+
+#include "../../../../lib/vdso/getrandom.c"
+
+ssize_t __vdso_getrandom(void *buffer, size_t len, unsigned int flags, void *state);
+
+ssize_t __vdso_getrandom(void *buffer, size_t len, unsigned int flags, void *state)
+{
+	return __cvdso_getrandom(buffer, len, flags, state);
+}
+
+ssize_t getrandom(void *, size_t, unsigned int, void *)
+	__attribute__((weak, alias("__vdso_getrandom")));
arch/x86/include/asm/vdso/getrandom.h new file +55 -0

55 lines not on this page · raw message

arch/x86/include/asm/vdso/vsyscall.h +2 -0
index be199a9b2676..71c56586a22f 100644
@@ -11,6 +11,8 @@
 #include <asm/vvar.h>
 
 DEFINE_VVAR(struct vdso_data, _vdso_data);
+DEFINE_VVAR_SINGLE(struct vdso_rng_data, _vdso_rng_data);
+
 /*
  * Update the vDSO data page to keep in sync with kernel timekeeping.
  */
arch/x86/include/asm/vvar.h +16 -0
index 183e98e49ab9..9d9af37f7cab 100644
@@ -26,6 +26,8 @@
  */
 #define DECLARE_VVAR(offset, type, name) \
 	EMIT_VVAR(name, offset)
+#define DECLARE_VVAR_SINGLE(offset, type, name) \
+	EMIT_VVAR(name, offset)
 
 #else
 
@@ -37,6 +39,10 @@ extern char __vvar_page;
 	extern type timens_ ## name[CS_BASES]				\
 	__attribute__((visibility("hidden")));				\
 
+#define DECLARE_VVAR_SINGLE(offset, type, name)				\
+	extern type vvar_ ## name					\
+	__attribute__((visibility("hidden")));				\
+
 #define VVAR(name) (vvar_ ## name)
 #define TIMENS(name) (timens_ ## name)
 
@@ -44,12 +50,22 @@ extern char __vvar_page;
 	type name[CS_BASES]						\
 	__attribute__((section(".vvar_" #name), aligned(16))) __visible
 
+#define DEFINE_VVAR_SINGLE(type, name)					\
+	type name							\
+	__attribute__((section(".vvar_" #name), aligned(16))) __visible
+
 #endif
 
 /* DECLARE_VVAR(offset, type, name) */
 
 DECLARE_VVAR(128, struct vdso_data, _vdso_data)
 
+#if !defined(_SINGLE_DATA)
+#define _SINGLE_DATA
+DECLARE_VVAR_SINGLE(640, struct vdso_rng_data, _vdso_rng_data)
+#endif
+
 #undef DECLARE_VVAR
+#undef DECLARE_VVAR_SINGLE
 
 #endif
Signature · 1 line
2.38.1
#33 lore raw

Re: [PATCH v10 4/4] x86: vdso: Wire up getrandom() vDSO implementation

Quoted 9 lines · Jason A. Donenfeld

Hook up the generic vDSO implementation to the x86 vDSO data page. Since the existing vDSO infrastructure is heavily based on the timekeeping functionality, which works over arrays of bases, a new macro is introduced for vvars that are not arrays.

The vDSO function requires a ChaCha20 implementation that does not write to the stack, yet can still do an entire ChaCha20 permutation, so provide this using SSE2, since this is userland code that must work on all x86-64 processors.

Way more consumable and looks about right. Please take your time and give others a chance to look at this lot before rushing out v11.

Thanks,

tglx

author maintainer
#34 lore raw

Re: [PATCH v10 4/4] x86: vdso: Wire up getrandom() vDSO implementation

Hi Thomas,

Quoted 6 lines · Thomas Gleixner
Quoted 5 lines · Jason A. Donenfeld

Hook up the generic vDSO implementation to the x86 vDSO data page. Since the existing vDSO infrastructure is heavily based on the timekeeping functionality, which works over arrays of bases, a new macro is introduced for vvars that are not arrays.

Quoted 6 lines
Quoted 4 lines

The vDSO function requires a ChaCha20 implementation that does not write to the stack, yet can still do an entire ChaCha20 permutation, so provide this using SSE2, since this is userland code that must work on all x86-64 processors.

Way more consumable and looks about right.

Good.

Quoted 2 lines

Please take your time and
give others a chance to look at this lot before rushing out v11.

That's my plan indeed. Now that the patch is reviewable, I'll let it sit for a while. In between v10 and v11, my scratch work will be in
<https://git.zx2c4.com/linux-rng/log/?h=vdso>, which is rebased often.

Thanks again for looking this over.

Jason

#35 lore raw

Re: [PATCH v10 4/4] x86: vdso: Wire up getrandom() vDSO implementation

In reply to Jason A. Donenfeld
Quoted 20 lines · Jason A. Donenfeld
arch/x86/entry/vdso/vgetrandom-chacha.S new file +14 -0
new file mode 100644
index 000000000000..91fbb7ac7af4
@@ -0,0 +1,177 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2022 Jason A. Donenfeld <[email protected]>. All Rights Reserved.
+ */
+
+#include <linux/linkage.h>
+#include <asm/frame.h>
+
+.section	.rodata.cst16.CONSTANTS, "aM", @progbits, 16
+.align 16
+CONSTANTS:	.octa 0x6b20657479622d323320646e61707865
+.text
+
+/*

arch/x86/entry/vdso/vgetrandom-chacha.S line 24

Quoted 10 lines
arch/x86/entry/vdso/vgetrandom-chacha.S new file +10 -0
+ * Very basic SSE2 implementation of ChaCha20. Produces a given positive number
+ * of blocks of output with a nonce of 0, taking an input key and 8-byte
+ * counter. Importantly does not spill to the stack. Its arguments are:
+ *
+ *	rdi: output bytes
+ *	rsi: 32-byte key input
+ *	rdx: 8-byte counter input/output
+ *	rcx: number of 64-byte blocks to write to output
+ */
+SYM_FUNC_START(__arch_chacha20_blocks_nostack)

How was this ChaCha20 implementation tested?

It really ought to have some sort of test.

Wouldn't this be a good candidate for a KUnit test?

- Eric

author maintainer
#36 lore raw

Re: [PATCH v10 4/4] x86: vdso: Wire up getrandom() vDSO implementation

Hi Eric,

Quoted 34 lines · Eric Biggers
Quoted 20 lines · Jason A. Donenfeld
arch/x86/entry/vdso/vgetrandom-chacha.S new file +14 -0
new file mode 100644
index 000000000000..91fbb7ac7af4
@@ -0,0 +1,177 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2022 Jason A. Donenfeld <[email protected]>. All Rights Reserved.
+ */
+
+#include <linux/linkage.h>
+#include <asm/frame.h>
+
+.section	.rodata.cst16.CONSTANTS, "aM", @progbits, 16
+.align 16
+CONSTANTS:	.octa 0x6b20657479622d323320646e61707865
+.text
+
+/*

arch/x86/entry/vdso/vgetrandom-chacha.S line 24

Quoted 10 lines
arch/x86/entry/vdso/vgetrandom-chacha.S new file +10 -0
+ * Very basic SSE2 implementation of ChaCha20. Produces a given positive number
+ * of blocks of output with a nonce of 0, taking an input key and 8-byte
+ * counter. Importantly does not spill to the stack. Its arguments are:
+ *
+ *	rdi: output bytes
+ *	rsi: 32-byte key input
+ *	rdx: 8-byte counter input/output
+ *	rcx: number of 64-byte blocks to write to output
+ */
+SYM_FUNC_START(__arch_chacha20_blocks_nostack)

How was this ChaCha20 implementation tested?

Quoted 1 line

It really ought to have some sort of test.

I've been comparing different output lengths with what libsodium produces. ARX, so no bigint stuff with carry bugs or whatever. I'll see if I can make a good test to add to one of the various suites for v11.

Jason