On little-endian systems, doing subtraction after htons() leads to interesting results:
Given: MAGIC_BYTES = 123 = 0x007B aka. in big endian: 0x7B00 = 31488 sizeof(struct iphdr) = 20
Before this patch: __bpf_constant_htons(MAGIC_BYTES) - sizeof(struct iphdr) = 0x7AEC 0x7AEC = htons(0xEC7A) = htons(60538)
So these were outer IP packets with a total length of 123 bytes, containing an inner IP packet with a total length of 60538 bytes.
After this patch: __bpf_constant_htons(MAGIC_BYTES - sizeof(struct iphdr)) = htons(103)
Now these packets are outer IP packets with a total length of 123 bytes, containing an inner IP packet with a total length of 103 bytes.
Signed-off-by: Asbjørn Sloth Tønnesen ast@fiberby.net --- I didn't target bpf and add a Fixes: e853ae776a58 ("selftests/bpf: support BPF_FLOW_DISSECTOR_F_STOP_AT_ENCAP"), since it only breaks when I change the BPF flow dissector to interact with tot_len.
.../selftests/bpf/prog_tests/flow_dissector.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/tools/testing/selftests/bpf/prog_tests/flow_dissector.c b/tools/testing/selftests/bpf/prog_tests/flow_dissector.c index 9e5f38739104..6b3078dd5645 100644 --- a/tools/testing/selftests/bpf/prog_tests/flow_dissector.c +++ b/tools/testing/selftests/bpf/prog_tests/flow_dissector.c @@ -378,8 +378,8 @@ struct test tests[] = { .iph_inner.ihl = 5, .iph_inner.protocol = IPPROTO_TCP, .iph_inner.tot_len = - __bpf_constant_htons(MAGIC_BYTES) - - sizeof(struct iphdr), + __bpf_constant_htons(MAGIC_BYTES - + sizeof(struct iphdr)), .tcp.doff = 5, .tcp.source = 80, .tcp.dest = 8080, @@ -407,8 +407,8 @@ struct test tests[] = { .iph_inner.ihl = 5, .iph_inner.protocol = IPPROTO_TCP, .iph_inner.tot_len = - __bpf_constant_htons(MAGIC_BYTES) - - sizeof(struct iphdr), + __bpf_constant_htons(MAGIC_BYTES - + sizeof(struct iphdr)), .tcp.doff = 5, .tcp.source = 80, .tcp.dest = 8080, @@ -436,8 +436,8 @@ struct test tests[] = { .iph_inner.ihl = 5, .iph_inner.protocol = IPPROTO_TCP, .iph_inner.tot_len = - __bpf_constant_htons(MAGIC_BYTES) - - sizeof(struct iphdr), + __bpf_constant_htons(MAGIC_BYTES - + sizeof(struct iphdr)), .tcp.doff = 5, .tcp.source = 99, .tcp.dest = 9090,
Asbjørn Sloth Tønnesen ast@fiberby.net writes:
On little-endian systems, doing subtraction after htons() leads to interesting results:
Given: MAGIC_BYTES = 123 = 0x007B aka. in big endian: 0x7B00 = 31488 sizeof(struct iphdr) = 20
Before this patch: __bpf_constant_htons(MAGIC_BYTES) - sizeof(struct iphdr) = 0x7AEC 0x7AEC = htons(0xEC7A) = htons(60538)
So these were outer IP packets with a total length of 123 bytes, containing an inner IP packet with a total length of 60538 bytes.
It's just using bag of holding technology!
After this patch: __bpf_constant_htons(MAGIC_BYTES - sizeof(struct iphdr)) = htons(103)
Now these packets are outer IP packets with a total length of 123 bytes, containing an inner IP packet with a total length of 103 bytes.
Signed-off-by: Asbjørn Sloth Tønnesen ast@fiberby.net
Reviewed-by: Toke Høiland-Jørgensen toke@kernel.org
On 8/8/24 2:24 AM, Toke Høiland-Jørgensen wrote:
Asbjørn Sloth Tønnesen ast@fiberby.net writes:
On little-endian systems, doing subtraction after htons() leads to interesting results:
Given: MAGIC_BYTES = 123 = 0x007B aka. in big endian: 0x7B00 = 31488 sizeof(struct iphdr) = 20
Before this patch: __bpf_constant_htons(MAGIC_BYTES) - sizeof(struct iphdr) = 0x7AEC 0x7AEC = htons(0xEC7A) = htons(60538)
So these were outer IP packets with a total length of 123 bytes, containing an inner IP packet with a total length of 60538 bytes.
It's just using bag of holding technology!
After this patch: __bpf_constant_htons(MAGIC_BYTES - sizeof(struct iphdr)) = htons(103)
Now these packets are outer IP packets with a total length of 123 bytes, containing an inner IP packet with a total length of 103 bytes.
Signed-off-by: Asbjørn Sloth Tønnesen ast@fiberby.net
Reviewed-by: Toke Høiland-Jørgensen toke@kernel.org
Applied to bpf-next/net. Thanks.
linux-kselftest-mirror@lists.linaro.org