Skip to contentSkip to conversation

bpf: Add a generic bits iterator

PATCH Superseded Yafang Shao 2 participants 11 messages May 6, 2024 – May 9, 2024

Subsystems (3)

v7 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#0 lore raw

v7 cover bpf: Add a generic bits iterator

hree new kfuncs, namely bpf_iter_bits_{new,next,destroy}, have been added for the new bpf_iter_bits functionality. These kfuncs enable the iteration of the bits from a given address and a given number of bits.

- 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 to be iterated over is (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 can be used in any context and on any address.

Changes:
- v6->v7:
- Fix endianness error for non-long-aligned data (Andrii)
- v5->v6:
- Add positive tests (Andrii)
- v4->v5:
- Simplify test cases (Andrii)
- v3->v4:
- Fix endianness error on s390x (Andrii)
- zero-initialize kit->bits_copy and zero out nr_bits (Andrii)
- v2->v3:
- Optimization for u64/u32 mask (Andrii)
- v1->v2:
- Simplify the CPU number verification code to avoid the failure on s390x (Eduard)
- bpf: Add bpf_iter_cpumask
https://lwn.net/Articles/961104/
- bpf: Add new bpf helper bpf_for_each_cpu
https://lwn.net/Articles/939939/

Yafang Shao (2):
bpf: Add bits iterator
selftests/bpf: Add selftest for bits iter

 kernel/bpf/helpers.c                          | 140 +++++++++++++++
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 3 files changed, 302 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
Signature · 1 line
2.30.1 (Apple Git-130)
#1 lore raw

PATCH v7 1/2 bpf: Add bits iterator

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.

---
 kernel/bpf/helpers.c | 140 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 140 insertions(+)
kernel/bpf/helpers.c +140 -0
index 2a69a9a36c0f..83b2a02f795f 100644
@@ -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
+ * @nr_bits: The number of bits to be iterated over. Due to the limitation of
+ * memalloc, it can't greater than (4096 * 8).
+ *
+ * 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
+ * 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;
+
+		err = bpf_probe_read_kernel_common(((char *)&kit->bits_copy) + offset,
+						   size, unsafe_ptr__ign);
+		if (err)
+			return -EFAULT;
+
+		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;
+	}
+
+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 = {
Signature · 1 line
2.30.1 (Apple Git-130)
#2 lore raw

Re: [PATCH v7 bpf-next 1/2] bpf: Add bits iterator

Quoted 34 lines · Yafang Shao

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.

---
 kernel/bpf/helpers.c | 140 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 140 insertions(+)
kernel/bpf/helpers.c +7 -0
index 2a69a9a36c0f..83b2a02f795f 100644
@@ -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;

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2763

Quoted 10 lines
kernel/bpf/helpers.c +10 -0
+               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

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2765

Quoted 2 lines
kernel/bpf/helpers.c +2 -0
+ * @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

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2769

Quoted 4 lines
kernel/bpf/helpers.c +4 -0
+ *
+ * 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/

Quoted 18 lines
kernel/bpf/helpers.c +18 -0
+ * 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) {

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2797

Quoted 10 lines
kernel/bpf/helpers.c +10 -0
+               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

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2802

Quoted 5 lines
kernel/bpf/helpers.c +5 -0
+
+               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).

pw-bot: cr

Quoted 19 lines
kernel/bpf/helpers.c +19 -0
+
+               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);

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2831

Quoted 10 lines
kernel/bpf/helpers.c +10 -0
+       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.

Quoted 68 lines
kernel/bpf/helpers.c +55 -1
+
+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)

#3 lore raw

Re: [PATCH v7 bpf-next 1/2] bpf: Add bits iterator

Quoted 40 lines · Andrii Nakryiko
Quoted 38 lines · Yafang Shao

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.

---
 kernel/bpf/helpers.c | 140 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 140 insertions(+)
kernel/bpf/helpers.c +11 -0
index 2a69a9a36c0f..83b2a02f795f 100644
@@ -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;
Quoted 8 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2763

Quoted 6 lines
kernel/bpf/helpers.c +6 -0
+} __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.

Quoted 44 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2765

Quoted 3 lines
kernel/bpf/helpers.c +2 -0
+ * @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

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2769

Quoted 4 lines
kernel/bpf/helpers.c +4 -0
+ *
+ * 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/

Quoted 18 lines
kernel/bpf/helpers.c +18 -0
+ * 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) {

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2797

Quoted 10 lines
kernel/bpf/helpers.c +10 -0
+               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

Quoted 2 lines

there is __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ check throughout the kernel to do this detection

I missed that. will check it.

Quoted 24 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2802

Quoted 6 lines
kernel/bpf/helpers.c +5 -0
+
+               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;

Quoted 5 lines

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.

Quoted 31 lines

pw-bot: cr

Quoted 27 lines
kernel/bpf/helpers.c +27 -0
+
+               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);
Quoted 8 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2831

Quoted 2 lines
kernel/bpf/helpers.c +2 -0
+               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.

Quoted 69 lines
Quoted 69 lines
kernel/bpf/helpers.c +55 -1
+
+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)

Signature · 2 lines
Regards
Yafang
#4 lore raw

Re: [PATCH v7 bpf-next 1/2] bpf: Add bits iterator

Quoted 175 lines · Yafang Shao
Quoted 40 lines · Andrii Nakryiko
Quoted 38 lines · Yafang Shao

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.

---
 kernel/bpf/helpers.c | 140 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 140 insertions(+)
kernel/bpf/helpers.c +11 -0
index 2a69a9a36c0f..83b2a02f795f 100644
@@ -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;
Quoted 8 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2763

Quoted 6 lines
kernel/bpf/helpers.c +6 -0
+} __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.

Quoted 44 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2765

Quoted 3 lines
kernel/bpf/helpers.c +2 -0
+ * @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

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2769

Quoted 4 lines
kernel/bpf/helpers.c +4 -0
+ *
+ * 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/

Quoted 18 lines
kernel/bpf/helpers.c +18 -0
+ * 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) {

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2797

Quoted 10 lines
kernel/bpf/helpers.c +10 -0
+               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

Quoted 2 lines

there is __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ check throughout the kernel to do this detection

I missed that. will check it.

Quoted 24 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2802

Quoted 6 lines
kernel/bpf/helpers.c +5 -0
+
+               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;

Quoted 5 lines

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.

Quoted 31 lines

pw-bot: cr

Quoted 27 lines
kernel/bpf/helpers.c +27 -0
+
+               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);
Quoted 8 lines

kernel/bpf/helpers.c __bpf_kfunc void bpf_preempt_enable(void) · line 2831

Quoted 2 lines
kernel/bpf/helpers.c +2 -0
+               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.

Quoted 5 lines

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.

Right, but the question is whether this iterator should make the same simplifying assumption or not? I think the motivation for this iterator was the ability to iterate over CPU masks, so I'm asking (and that's why I cc'ed David) what we should do to make it work well for CPU masks.

Quoted 75 lines
Quoted 65 lines
Quoted 65 lines
kernel/bpf/helpers.c +55 -0
+
+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)
 
Quoted 4 lines
Quoted 4 lines
kernel/bpf/helpers.c +0 -1
 static const struct btf_kfunc_id_set common_kfunc_set = {
--

2.30.1 (Apple Git-130)

--
Regards
Yafang

#5 lore raw

PATCH v7 2/2 selftests/bpf: Add selftest for bits iter

In reply to Yafang Shao

Add test cases for the bits iter:
- positive case
- bit mask smaller than 8 bytes
- a typical case of having 8-byte bit mask
- another typical case where bit mask is > 8 bytes
- the index of set bit

- nagative cases
- bpf_iter_bits_destroy() is required after calling

    bpf_iter_bits_new()

- bpf_iter_bits_destroy() can only destroy an initialized iter
- bpf_iter_bits_next() must use an initialized iter

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 2 files changed, 162 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index c4f9f306646e..7e04ecaaa20a 100644
@@ -84,6 +84,7 @@
 #include "verifier_xadd.skel.h"
 #include "verifier_xdp.skel.h"
 #include "verifier_xdp_direct_packet_access.skel.h"
+#include "verifier_bits_iter.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -198,6 +199,7 @@ void test_verifier_var_off(void)              { RUN(verifier_var_off); }
 void test_verifier_xadd(void)                 { RUN(verifier_xadd); }
 void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
+void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 
 static int init_test_val_map(struct bpf_object *obj, char *map_name)
 {
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +160 -0

160 lines not on this page · raw message

Signature · 1 line
2.30.1 (Apple Git-130)
#6 lore raw

Re: [PATCH v7 bpf-next 2/2] selftests/bpf: Add selftest for bits iter

Quoted 102 lines · Yafang Shao

Add test cases for the bits iter:
- positive case
- bit mask smaller than 8 bytes
- a typical case of having 8-byte bit mask
- another typical case where bit mask is > 8 bytes
- the index of set bit

- nagative cases
- bpf_iter_bits_destroy() is required after calling

    bpf_iter_bits_new()

- bpf_iter_bits_destroy() can only destroy an initialized iter
- bpf_iter_bits_next() must use an initialized iter

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 2 files changed, 162 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index c4f9f306646e..7e04ecaaa20a 100644
@@ -84,6 +84,7 @@
 #include "verifier_xadd.skel.h"
 #include "verifier_xdp.skel.h"
 #include "verifier_xdp_direct_packet_access.skel.h"
+#include "verifier_bits_iter.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -198,6 +199,7 @@ void test_verifier_var_off(void)              { RUN(verifier_var_off); }
 void test_verifier_xadd(void)                 { RUN(verifier_xadd); }
 void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
+void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 
 static int init_test_val_map(struct bpf_object *obj, char *map_name)
 {
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +55 -0
new file mode 100644
index 000000000000..2f7b62b25638
@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[email protected]> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "bpf_misc.h"
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+                     u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+SEC("iter.s/cgroup")
+__description("bits iter without destroy")
+__failure __msg("Unreleased reference")
+int BPF_PROG(no_destroy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it;
+       struct task_struct *p;
+
+       p = bpf_task_from_pid(1);
+       if (!p)
+               return 1;
+
+       bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+       bpf_iter_bits_next(&it);
+       bpf_task_release(p);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->next()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(next_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits *it = NULL;
+
+       bpf_iter_bits_next(it);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->destroy()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(destroy_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it = {};
+
+       bpf_iter_bits_destroy(&it);

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 65

Quoted 10 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +10 -0
+       return 0;
+}
+
+SEC("syscall")
+__description("bits copy 32")
+__success __retval(10)
+int bits_copy32(void)
+{
+       /* 21 bits:             --------------------- */
+       u32 data = 0b11111101111101111100001000100101U;

if you define this bit mask as an array of bytes, then you won't have to handle big-endian in the tests at all

Quoted 16 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +16 -0
+       int nr = 0, offset = 0;
+       int *bit;
+
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u32) - (21 + 7) / 8;
+#endif
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 21)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits copy 64")
+__success __retval(18)
+int bits_copy64(void)
+{

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 91

Quoted 10 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +10 -0
+       /* 34 bits:         ~-------- */
+       u64 data = 0xffffefdf0f0f0f0fUL;
+       int nr = 0, offset = 0;
+       int *bit;
+
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u64) - (34 + 7) / 8;
+#endif
+
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 34)

see above about byte array, but if we define different (not as byte array but long[]), it would be cleaner to have

#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
u64 data = 0x......UL;
#else
u64 data = 0x......UL;
#endif

wherer we'd hard-code bit masks in proper endianness in one place and then just do a clean bpf_for_each(bits, bit, &data, <len>) {} calls

Quoted 72 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +69 -1
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc long-aligned")
+__success __retval(32) /* 16 * 2 */
+int bits_memalloc(void)
+{
+       char data[16];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x48, sizeof(data));
+       bpf_for_each(bits, bit, &data, sizeof(data) * 8)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc non-long-aligned")
+__success __retval(85) /* 17 * 5*/
+int bits_memalloc_non_aligned(void)
+{
+       char data[17];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x1f, sizeof(data));
+       bpf_for_each(bits, bit, &data, sizeof(data) * 8)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc non-aligned-bits")
+__success __retval(27) /* 8 * 3 + 3 */
+int bits_memalloc_non_aligned_bits(void)
+{
+       char data[16];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x31, sizeof(data));
+       /* Different with all other bytes */
+       data[8] = 0xf7;
+
+       bpf_for_each(bits, bit, &data,  68)
+               nr++;
+       return nr;
+}
+
+
+SEC("syscall")
+__description("bit index")
+__success __retval(8)
+int bit_index(void)
+{
+       u64 data = 0x100;
+       int bit_idx = 0;
+       int *bit;
+
+       bpf_for_each(bits, bit, &data, 64) {
+               if (*bit == 0)
+                       continue;
+               bit_idx = *bit;
+       }
+       return bit_idx;
+}
--

2.30.1 (Apple Git-130)

#7 lore raw

Re: [PATCH v7 bpf-next 2/2] selftests/bpf: Add selftest for bits iter

Quoted 109 lines · Andrii Nakryiko
Quoted 107 lines · Yafang Shao

Add test cases for the bits iter:
- positive case
- bit mask smaller than 8 bytes
- a typical case of having 8-byte bit mask
- another typical case where bit mask is > 8 bytes
- the index of set bit

- nagative cases
- bpf_iter_bits_destroy() is required after calling

    bpf_iter_bits_new()

- bpf_iter_bits_destroy() can only destroy an initialized iter
- bpf_iter_bits_next() must use an initialized iter

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 2 files changed, 162 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index c4f9f306646e..7e04ecaaa20a 100644
@@ -84,6 +84,7 @@
 #include "verifier_xadd.skel.h"
 #include "verifier_xdp.skel.h"
 #include "verifier_xdp_direct_packet_access.skel.h"
+#include "verifier_bits_iter.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -198,6 +199,7 @@ void test_verifier_var_off(void)              { RUN(verifier_var_off); }
 void test_verifier_xadd(void)                 { RUN(verifier_xadd); }
 void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
+void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 
 static int init_test_val_map(struct bpf_object *obj, char *map_name)
 {
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +60 -0
new file mode 100644
index 000000000000..2f7b62b25638
@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[email protected]> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "bpf_misc.h"
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+                     u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+SEC("iter.s/cgroup")
+__description("bits iter without destroy")
+__failure __msg("Unreleased reference")
+int BPF_PROG(no_destroy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it;
+       struct task_struct *p;
+
+       p = bpf_task_from_pid(1);
+       if (!p)
+               return 1;
+
+       bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+       bpf_iter_bits_next(&it);
+       bpf_task_release(p);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->next()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(next_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits *it = NULL;
+
+       bpf_iter_bits_next(it);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->destroy()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(destroy_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it = {};
+
+       bpf_iter_bits_destroy(&it);
+       return 0;
+}
+
+SEC("syscall")
+__description("bits copy 32")
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 65

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+__success __retval(10)
+int bits_copy32(void)
+{
+       /* 21 bits:             --------------------- */
+       u32 data = 0b11111101111101111100001000100101U;

if you define this bit mask as an array of bytes, then you won't have to handle big-endian in the tests at all

This test case provides a clear example of iterating over data of type u32, offering valuable guidance for users who need to perform such iterations.

Quoted 23 lines
Quoted 23 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +21 -0
+       int nr = 0, offset = 0;
+       int *bit;
+
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u32) - (21 + 7) / 8;
+#endif
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 21)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits copy 64")
+__success __retval(18)
+int bits_copy64(void)
+{
+       /* 34 bits:         ~-------- */
+       u64 data = 0xffffefdf0f0f0f0fUL;
+       int nr = 0, offset = 0;
+       int *bit;
+
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 91

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u64) - (34 + 7) / 8;
+#endif
+
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 34)

see above about byte array, but if we define different (not as byte array but long[]), it would be cleaner to have

This test case demonstrates how to iterate over data of type u64.

Quoted 6 lines

#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
u64 data = 0x......UL;
#else
u64 data = 0x......UL;
#endif

looks good.

Quoted 76 lines

wherer we'd hard-code bit masks in proper endianness in one place and then just do a clean bpf_for_each(bits, bit, &data, <len>) {} calls

Quoted 72 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +69 -1
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc long-aligned")
+__success __retval(32) /* 16 * 2 */
+int bits_memalloc(void)
+{
+       char data[16];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x48, sizeof(data));
+       bpf_for_each(bits, bit, &data, sizeof(data) * 8)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc non-long-aligned")
+__success __retval(85) /* 17 * 5*/
+int bits_memalloc_non_aligned(void)
+{
+       char data[17];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x1f, sizeof(data));
+       bpf_for_each(bits, bit, &data, sizeof(data) * 8)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc non-aligned-bits")
+__success __retval(27) /* 8 * 3 + 3 */
+int bits_memalloc_non_aligned_bits(void)
+{
+       char data[16];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x31, sizeof(data));
+       /* Different with all other bytes */
+       data[8] = 0xf7;
+
+       bpf_for_each(bits, bit, &data,  68)
+               nr++;
+       return nr;
+}
+
+
+SEC("syscall")
+__description("bit index")
+__success __retval(8)
+int bit_index(void)
+{
+       u64 data = 0x100;
+       int bit_idx = 0;
+       int *bit;
+
+       bpf_for_each(bits, bit, &data, 64) {
+               if (*bit == 0)
+                       continue;
+               bit_idx = *bit;
+       }
+       return bit_idx;
+}
--

2.30.1 (Apple Git-130)

Signature · 2 lines
Regards
Yafang
#8 lore raw

Re: [PATCH v7 bpf-next 2/2] selftests/bpf: Add selftest for bits iter

Quoted 160 lines · Yafang Shao
Quoted 109 lines · Andrii Nakryiko
Quoted 107 lines · Yafang Shao

Add test cases for the bits iter:
- positive case
- bit mask smaller than 8 bytes
- a typical case of having 8-byte bit mask
- another typical case where bit mask is > 8 bytes
- the index of set bit

- nagative cases
- bpf_iter_bits_destroy() is required after calling

    bpf_iter_bits_new()

- bpf_iter_bits_destroy() can only destroy an initialized iter
- bpf_iter_bits_next() must use an initialized iter

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 2 files changed, 162 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index c4f9f306646e..7e04ecaaa20a 100644
@@ -84,6 +84,7 @@
 #include "verifier_xadd.skel.h"
 #include "verifier_xdp.skel.h"
 #include "verifier_xdp_direct_packet_access.skel.h"
+#include "verifier_bits_iter.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -198,6 +199,7 @@ void test_verifier_var_off(void)              { RUN(verifier_var_off); }
 void test_verifier_xadd(void)                 { RUN(verifier_xadd); }
 void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
+void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 
 static int init_test_val_map(struct bpf_object *obj, char *map_name)
 {
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +60 -0
new file mode 100644
index 000000000000..2f7b62b25638
@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[email protected]> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "bpf_misc.h"
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+                     u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+SEC("iter.s/cgroup")
+__description("bits iter without destroy")
+__failure __msg("Unreleased reference")
+int BPF_PROG(no_destroy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it;
+       struct task_struct *p;
+
+       p = bpf_task_from_pid(1);
+       if (!p)
+               return 1;
+
+       bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+       bpf_iter_bits_next(&it);
+       bpf_task_release(p);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->next()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(next_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits *it = NULL;
+
+       bpf_iter_bits_next(it);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->destroy()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(destroy_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it = {};
+
+       bpf_iter_bits_destroy(&it);
+       return 0;
+}
+
+SEC("syscall")
+__description("bits copy 32")
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 65

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+__success __retval(10)
+int bits_copy32(void)
+{
+       /* 21 bits:             --------------------- */
+       u32 data = 0b11111101111101111100001000100101U;

if you define this bit mask as an array of bytes, then you won't have to handle big-endian in the tests at all

This test case provides a clear example of iterating over data of type u32, offering valuable guidance for users who need to perform such iterations.

Quoted 23 lines
Quoted 23 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +21 -0
+       int nr = 0, offset = 0;
+       int *bit;
+
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u32) - (21 + 7) / 8;
+#endif
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 21)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits copy 64")
+__success __retval(18)
+int bits_copy64(void)
+{
+       /* 34 bits:         ~-------- */
+       u64 data = 0xffffefdf0f0f0f0fUL;
+       int nr = 0, offset = 0;
+       int *bit;
+
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 91

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u64) - (34 + 7) / 8;
+#endif
+
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 34)

see above about byte array, but if we define different (not as byte array but long[]), it would be cleaner to have

This test case demonstrates how to iterate over data of type u64.

Quoted 1 line
Quoted 8 lines
Quoted 5 lines

#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
u64 data = 0x......UL;
#else
u64 data = 0x......UL;
#endif

looks good.

Please hold off on sending a new revision until we figure out what the contract should be. Because I feel like it's a (relatively) big decision whether a bit mask is treated as an array of bytes or as an array of longs. For little-endian it makes no difference, but for big-endian it's a big difference and has usability and performance implications.

Quoted 82 lines
Quoted 68 lines

wherer we'd hard-code bit masks in proper endianness in one place and then just do a clean bpf_for_each(bits, bit, &data, <len>) {} calls

Quoted 64 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +64 -0
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc long-aligned")
+__success __retval(32) /* 16 * 2 */
+int bits_memalloc(void)
+{
+       char data[16];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x48, sizeof(data));
+       bpf_for_each(bits, bit, &data, sizeof(data) * 8)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc non-long-aligned")
+__success __retval(85) /* 17 * 5*/
+int bits_memalloc_non_aligned(void)
+{
+       char data[17];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x1f, sizeof(data));
+       bpf_for_each(bits, bit, &data, sizeof(data) * 8)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits memalloc non-aligned-bits")
+__success __retval(27) /* 8 * 3 + 3 */
+int bits_memalloc_non_aligned_bits(void)
+{
+       char data[16];
+       int nr = 0;
+       int *bit;
+
+       __builtin_memset(&data, 0x31, sizeof(data));
+       /* Different with all other bytes */
+       data[8] = 0xf7;
+
+       bpf_for_each(bits, bit, &data,  68)
+               nr++;
+       return nr;
+}
+
+
+SEC("syscall")
+__description("bit index")
+__success __retval(8)
+int bit_index(void)
+{
+       u64 data = 0x100;
+       int bit_idx = 0;
+       int *bit;
+
+       bpf_for_each(bits, bit, &data, 64) {
+               if (*bit == 0)
Quoted 8 lines
Quoted 8 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -1
+                       continue;
+               bit_idx = *bit;
+       }
+       return bit_idx;
+}
--

2.30.1 (Apple Git-130)

--
Regards
Yafang

#9 lore raw

Re: [PATCH v7 bpf-next 2/2] selftests/bpf: Add selftest for bits iter

Quoted 171 lines · Andrii Nakryiko
Quoted 160 lines · Yafang Shao
Quoted 109 lines · Andrii Nakryiko
Quoted 107 lines · Yafang Shao

Add test cases for the bits iter:
- positive case
- bit mask smaller than 8 bytes
- a typical case of having 8-byte bit mask
- another typical case where bit mask is > 8 bytes
- the index of set bit

- nagative cases
- bpf_iter_bits_destroy() is required after calling

    bpf_iter_bits_new()

- bpf_iter_bits_destroy() can only destroy an initialized iter
- bpf_iter_bits_next() must use an initialized iter

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 2 files changed, 162 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index c4f9f306646e..7e04ecaaa20a 100644
@@ -84,6 +84,7 @@
 #include "verifier_xadd.skel.h"
 #include "verifier_xdp.skel.h"
 #include "verifier_xdp_direct_packet_access.skel.h"
+#include "verifier_bits_iter.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -198,6 +199,7 @@ void test_verifier_var_off(void)              { RUN(verifier_var_off); }
 void test_verifier_xadd(void)                 { RUN(verifier_xadd); }
 void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
+void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 
 static int init_test_val_map(struct bpf_object *obj, char *map_name)
 {
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +60 -0
new file mode 100644
index 000000000000..2f7b62b25638
@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[email protected]> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "bpf_misc.h"
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+                     u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+SEC("iter.s/cgroup")
+__description("bits iter without destroy")
+__failure __msg("Unreleased reference")
+int BPF_PROG(no_destroy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it;
+       struct task_struct *p;
+
+       p = bpf_task_from_pid(1);
+       if (!p)
+               return 1;
+
+       bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+       bpf_iter_bits_next(&it);
+       bpf_task_release(p);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->next()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(next_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits *it = NULL;
+
+       bpf_iter_bits_next(it);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->destroy()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(destroy_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it = {};
+
+       bpf_iter_bits_destroy(&it);
+       return 0;
+}
+
+SEC("syscall")
+__description("bits copy 32")
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 65

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+__success __retval(10)
+int bits_copy32(void)
+{
+       /* 21 bits:             --------------------- */
+       u32 data = 0b11111101111101111100001000100101U;

if you define this bit mask as an array of bytes, then you won't have to handle big-endian in the tests at all

This test case provides a clear example of iterating over data of type u32, offering valuable guidance for users who need to perform such iterations.

Quoted 23 lines
Quoted 23 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +21 -0
+       int nr = 0, offset = 0;
+       int *bit;
+
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u32) - (21 + 7) / 8;
+#endif
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 21)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits copy 64")
+__success __retval(18)
+int bits_copy64(void)
+{
+       /* 34 bits:         ~-------- */
+       u64 data = 0xffffefdf0f0f0f0fUL;
+       int nr = 0, offset = 0;
+       int *bit;
+
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 91

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u64) - (34 + 7) / 8;
+#endif
+
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 34)

see above about byte array, but if we define different (not as byte array but long[]), it would be cleaner to have

This test case demonstrates how to iterate over data of type u64.

Quoted 1 line
Quoted 8 lines
Quoted 5 lines

#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
u64 data = 0x......UL;
#else
u64 data = 0x......UL;
#endif

looks good.

Quoted 6 lines

Please hold off on sending a new revision until we figure out what the contract should be. Because I feel like it's a (relatively) big decision whether a bit mask is treated as an array of bytes or as an array of longs. For little-endian it makes no difference, but for big-endian it's a big difference and has usability and performance implications.

Perhaps it would be advantageous to define the interface as follows:

bpf_iter_bits_new(struct bpf_iter_bits *it, const u64 *unsafe_ptr__ign, u32 words)

This approach eliminates the need to account for endianness.

Signature · 2 lines
Regards
Yafang
#10 lore raw

Re: [PATCH v7 bpf-next 2/2] selftests/bpf: Add selftest for bits iter

Quoted 181 lines · Yafang Shao
Quoted 171 lines · Andrii Nakryiko
Quoted 160 lines · Yafang Shao
Quoted 109 lines · Andrii Nakryiko
Quoted 107 lines · Yafang Shao

Add test cases for the bits iter:
- positive case
- bit mask smaller than 8 bytes
- a typical case of having 8-byte bit mask
- another typical case where bit mask is > 8 bytes
- the index of set bit

- nagative cases
- bpf_iter_bits_destroy() is required after calling

    bpf_iter_bits_new()

- bpf_iter_bits_destroy() can only destroy an initialized iter
- bpf_iter_bits_next() must use an initialized iter

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/verifier_bits_iter.c  | 160 ++++++++++++++++++
 2 files changed, 162 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/verifier_bits_iter.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index c4f9f306646e..7e04ecaaa20a 100644
@@ -84,6 +84,7 @@
 #include "verifier_xadd.skel.h"
 #include "verifier_xdp.skel.h"
 #include "verifier_xdp_direct_packet_access.skel.h"
+#include "verifier_bits_iter.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -198,6 +199,7 @@ void test_verifier_var_off(void)              { RUN(verifier_var_off); }
 void test_verifier_xadd(void)                 { RUN(verifier_xadd); }
 void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
+void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 
 static int init_test_val_map(struct bpf_object *obj, char *map_name)
 {
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +60 -0
new file mode 100644
index 000000000000..2f7b62b25638
@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[email protected]> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "bpf_misc.h"
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+                     u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+SEC("iter.s/cgroup")
+__description("bits iter without destroy")
+__failure __msg("Unreleased reference")
+int BPF_PROG(no_destroy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it;
+       struct task_struct *p;
+
+       p = bpf_task_from_pid(1);
+       if (!p)
+               return 1;
+
+       bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+       bpf_iter_bits_next(&it);
+       bpf_task_release(p);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->next()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(next_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits *it = NULL;
+
+       bpf_iter_bits_next(it);
+       return 0;
+}
+
+SEC("iter/cgroup")
+__description("bits iter with uninitialized iter in ->destroy()")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(destroy_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+       struct bpf_iter_bits it = {};
+
+       bpf_iter_bits_destroy(&it);
+       return 0;
+}
+
+SEC("syscall")
+__description("bits copy 32")
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 65

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+__success __retval(10)
+int bits_copy32(void)
+{
+       /* 21 bits:             --------------------- */
+       u32 data = 0b11111101111101111100001000100101U;

if you define this bit mask as an array of bytes, then you won't have to handle big-endian in the tests at all

This test case provides a clear example of iterating over data of type u32, offering valuable guidance for users who need to perform such iterations.

Quoted 23 lines
Quoted 23 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +21 -0
+       int nr = 0, offset = 0;
+       int *bit;
+
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u32) - (21 + 7) / 8;
+#endif
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 21)
+               nr++;
+       return nr;
+}
+
+SEC("syscall")
+__description("bits copy 64")
+__success __retval(18)
+int bits_copy64(void)
+{
+       /* 34 bits:         ~-------- */
+       u64 data = 0xffffefdf0f0f0f0fUL;
+       int nr = 0, offset = 0;
+       int *bit;
+
Quoted 8 lines

tools/testing/selftests/bpf/progs/verifier_bits_iter.c line 91

Quoted 5 lines
tools/testing/selftests/bpf/progs/verifier_bits_iter.c new file +5 -0
+#if defined(__TARGET_ARCH_s390)
+       offset = sizeof(u64) - (34 + 7) / 8;
+#endif
+
+       bpf_for_each(bits, bit, ((char *)&data) + offset, 34)

see above about byte array, but if we define different (not as byte array but long[]), it would be cleaner to have

This test case demonstrates how to iterate over data of type u64.

Quoted 1 line
Quoted 8 lines
Quoted 5 lines

#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
u64 data = 0x......UL;
#else
u64 data = 0x......UL;
#endif

looks good.

Quoted 6 lines

Please hold off on sending a new revision until we figure out what the contract should be. Because I feel like it's a (relatively) big decision whether a bit mask is treated as an array of bytes or as an array of longs. For little-endian it makes no difference, but for big-endian it's a big difference and has usability and performance implications.

Quoted 6 lines

Perhaps it would be advantageous to define the interface as follows:

bpf_iter_bits_new(struct bpf_iter_bits *it, const u64 *unsafe_ptr__ign, u32 words)

This approach eliminates the need to account for endianness.

I don't mind that, if others don't have any opinion. Let's just document that by "words" we mean 8-byte integers.

Quoted 4 lines

--
Regards
Yafang