On Tue, May 7, 2024 at 11:38 AM Andrii Nakryiko
<andrii.nakryiko@xxxxxxxxx> wrote:
>
> On Sun, May 5, 2024 at 8:35 PM Yafang Shao <laoar.shao@xxxxxxxxx> wrote:
> >
> > Add three new kfuncs for the bits iterator:
> > - bpf_iter_bits_new
> > Initialize a new bits iterator for a given memory area. Due to the
> > limitation of bpf memalloc, the max number of bits that can be iterated
> > over is limited to (4096 * 8).
> > - bpf_iter_bits_next
> > Get the next bit in a bpf_iter_bits
> > - bpf_iter_bits_destroy
> > Destroy a bpf_iter_bits
> >
> > The bits iterator facilitates the iteration of the bits of a memory area,
> > such as cpumask. It can be used in any context and on any address.
> >
> > Signed-off-by: Yafang Shao <laoar.shao@xxxxxxxxx>
> > ---
> > kernel/bpf/helpers.c | 140 +++++++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 140 insertions(+)
> >
> > diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> > index 2a69a9a36c0f..83b2a02f795f 100644
> > --- a/kernel/bpf/helpers.c
> > +++ b/kernel/bpf/helpers.c
> > @@ -2744,6 +2744,143 @@ __bpf_kfunc void bpf_preempt_enable(void)
> > preempt_enable();
> > }
> >
> > +struct bpf_iter_bits {
> > + __u64 __opaque[2];
> > +} __aligned(8);
> > +
> > +struct bpf_iter_bits_kern {
> > + union {
> > + unsigned long *bits;
> > + unsigned long bits_copy;
> > + };
> > + u32 nr_bits;
> > + int bit;
> > +} __aligned(8);
> > +
> > +/**
> > + * bpf_iter_bits_new() - Initialize a new bits iterator for a given memory area
> > + * @it: The new bpf_iter_bits to be created
> > + * @unsafe_ptr__ign: A ponter pointing to a memory area to be iterated over
>
> typo: pointer
Thanks for the fix and the other fixes.
>
> > + * @nr_bits: The number of bits to be iterated over. Due to the limitation of
> > + * memalloc, it can't greater than (4096 * 8).
>
> typo: can't be greater
>
> > + *
> > + * This function initializes a new bpf_iter_bits structure for iterating over
> > + * a memory area which is specified by the @unsafe_ptr__ign and @nr_bits. It
> > + * copy the data of the memory area to the newly created bpf_iter_bits @it for
>
> s/copy/copies/
>
> > + * subsequent iteration operations.
> > + *
> > + * On success, 0 is returned. On failure, ERR is returned.
> > + */
> > +__bpf_kfunc int
> > +bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign, u32 nr_bits)
> > +{
> > + struct bpf_iter_bits_kern *kit = (void *)it;
> > + u32 words = BITS_TO_LONGS(nr_bits);
> > + u32 size = BITS_TO_BYTES(nr_bits);
> > + u32 left, offset;
> > + int err;
> > +
> > + BUILD_BUG_ON(sizeof(struct bpf_iter_bits_kern) != sizeof(struct bpf_iter_bits));
> > + BUILD_BUG_ON(__alignof__(struct bpf_iter_bits_kern) !=
> > + __alignof__(struct bpf_iter_bits));
> > +
> > + if (!unsafe_ptr__ign || !nr_bits) {
> > + kit->bits = NULL;
> > + return -EINVAL;
> > + }
> > +
> > + kit->nr_bits = 0;
> > + kit->bits_copy = 0;
> > + /* Optimization for u64/u32 mask */
> > + if (nr_bits <= 64) {
> > + /* For big-endian, we must calculate the offset */
> > + offset = IS_ENABLED(CONFIG_S390) ? sizeof(u64) - size : 0;
>
> S390 isn't the only big-endian architecture, it's wrong to hard-code just S390
>
> there is __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ check throughout the
> kernel to do this detection
I missed that. will check it.
>
> > +
> > + err = bpf_probe_read_kernel_common(((char *)&kit->bits_copy) + offset,
> > + size, unsafe_ptr__ign);
> > + if (err)
> > + return -EFAULT;
>
> I'd rewrite the above to something like (not tested, but should give
> the right idea):
>
> long bits = 0;
>
> err = bpf_probe_read_kernel_common(&bits, size, unsafe_ptr__ign);
> if (err)
> return -EFAULT;
>
> #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
> bits = __swab64(bits);
> #endif
>
> /* deal with bit mask of weird size, ensuring upper bits are zero */
> bits <<= 64 - nr_bits;
> bits >>= 64 - nr_bits;
>
> kit->bits_copy = bits;
>
>
> This should take care of both big-endianness, and non-multiple-of-8
> sized bitmasks (I think, we need tests).
looks good, will change it.
>
> pw-bot: cr
>
>
> > +
> > + kit->nr_bits = nr_bits;
> > + kit->bit = -1;
> > + return 0;
> > + }
> > +
> > + /* Fallback to memalloc */
> > + kit->bits = bpf_mem_alloc(&bpf_global_ma, size);
> > + if (!kit->bits)
> > + return -ENOMEM;
> > +
> > + err = bpf_probe_read_kernel_common(kit->bits, words * sizeof(u64), unsafe_ptr__ign);
> > + if (err) {
> > + bpf_mem_free(&bpf_global_ma, kit->bits);
> > + return err;
> > + }
> > +
> > + /* long-aligned */
> > + left = size & (sizeof(u64) - 1);
> > + if (!left)
> > + goto out;
> > +
> > + offset = IS_ENABLED(CONFIG_S390) ? sizeof(u64) - left : 0;
> > + err = bpf_probe_read_kernel_common((char *)(kit->bits + words - 1) + offset, left,
> > + unsafe_ptr__ign + (words - 1) * sizeof(u64));
> > + if (err) {
> > + bpf_mem_free(&bpf_global_ma, kit->bits);
> > + return err;
> > + }
>
> tbh, I'm not sure what's the desired behavior here is. David (cc'ed),
> you were dealing with cpumasks, how is the bit mask specified there?
> Is it considered to be an long[] array or byte[] array? And how is
> that working on big-endian, because I think it makes a difference?
> Please take a look, thanks.
The function find_next_bit() requires the pointer to be of type
"unsigned long *", hence, we must ensure consistency by converting it
here as well. As cpumask represents a bitmap and is always of type
"unsigned long *", it remains unaffected by endianness considerations.
>
> > +
> > +out:
> > + kit->nr_bits = nr_bits;
> > + kit->bit = -1;
> > + return 0;
> > +}
> > +
> > +/**
> > + * bpf_iter_bits_next() - Get the next bit in a bpf_iter_bits
> > + * @it: The bpf_iter_bits to be checked
> > + *
> > + * This function returns a pointer to a number representing the value of the
> > + * next bit in the bits.
> > + *
> > + * If there are no further bit available, it returns NULL.
> > + */
> > +__bpf_kfunc int *bpf_iter_bits_next(struct bpf_iter_bits *it)
> > +{
> > + struct bpf_iter_bits_kern *kit = (void *)it;
> > + u32 nr_bits = kit->nr_bits;
> > + const unsigned long *bits;
> > + int bit;
> > +
> > + if (nr_bits == 0)
> > + return NULL;
> > +
> > + bits = nr_bits <= 64 ? &kit->bits_copy : kit->bits;
> > + bit = find_next_bit(bits, nr_bits, kit->bit + 1);
> > + if (bit >= nr_bits) {
> > + kit->nr_bits = 0;
> > + return NULL;
> > + }
> > +
> > + kit->bit = bit;
> > + return &kit->bit;
> > +}
> > +
> > +/**
> > + * bpf_iter_bits_destroy() - Destroy a bpf_iter_bits
> > + * @it: The bpf_iter_bits to be destroyed
> > + *
> > + * Destroy the resource associated with the bpf_iter_bits.
> > + */
> > +__bpf_kfunc void bpf_iter_bits_destroy(struct bpf_iter_bits *it)
> > +{
> > + struct bpf_iter_bits_kern *kit = (void *)it;
> > +
> > + if (kit->nr_bits <= 64)
> > + return;
> > + bpf_mem_free(&bpf_global_ma, kit->bits);
> > +}
> > +
> > __bpf_kfunc_end_defs();
> >
> > BTF_KFUNCS_START(generic_btf_ids)
> > @@ -2826,6 +2963,9 @@ BTF_ID_FLAGS(func, bpf_wq_set_callback_impl)
> > BTF_ID_FLAGS(func, bpf_wq_start)
> > BTF_ID_FLAGS(func, bpf_preempt_disable)
> > BTF_ID_FLAGS(func, bpf_preempt_enable)
> > +BTF_ID_FLAGS(func, bpf_iter_bits_new, KF_ITER_NEW)
> > +BTF_ID_FLAGS(func, bpf_iter_bits_next, KF_ITER_NEXT | KF_RET_NULL)
> > +BTF_ID_FLAGS(func, bpf_iter_bits_destroy, KF_ITER_DESTROY)
> > BTF_KFUNCS_END(common_btf_ids)
> >
> > static const struct btf_kfunc_id_set common_kfunc_set = {
> > --
> > 2.30.1 (Apple Git-130)
> >
--
Regards
Yafang
