Hi Stephen,
The new Kernel expects the BPF maps in BTF format:
https://www.kernel.org/doc/html/v5.4/bpf/btf.html.
So, we need to compile with "-g" option as follows:
==
clang -O2 -target bpf -c tap_bpf_program.c -g -o tap_bpf_program.o
=
I had cross check this functionality with DPDK-22.11 and as the BPF C-program
is same had modified just the C-program and submitted for review.
I will update the Makefile changes also and will publish for code-review once
again w.r.t DPDK-23.11.
Thanks,
Madhuker.
-Original Message-
From: Madhuker Mythri
Sent: 06 January 2024 11:06
To: Stephen Hemminger
Cc: ferruh.yi...@amd.com; dev@dpdk.org
Subject: RE: [External] : Re: [PATCH] net/tap: Modified TAP BPF program as per
the new Kernel-version upgrade requirements.
The flow creation worked well for us on the Azure/Hyper-V platform using
failsafe/tap PMD, using Kernel-5.15.0 and 5.4 versions.
So, does the original/existing code works well on Kernel-6.5 version ?
Thanks,
Madhuker.
-Original Message-
From: Stephen Hemminger
Sent: 06 January 2024 01:01
To: Madhuker Mythri
Cc: ferruh.yi...@amd.com; dev@dpdk.org
Subject: [External] : Re: [PATCH] net/tap: Modified TAP BPF program as per the
new Kernel-version upgrade requirements.
On Thu, 4 Jan 2024 22:57:56 +0530
madhuker.myt...@oracle.com wrote:
> adhuker Mythri
>
> When multiple queues configured, internally RSS will be enabled and thus TAP
> BPF RSS byte-code will be loaded on to the Kernel using BPF system calls.
>
> Here, the problem is loading the existing BPF byte-code to the Kernel-5.15
> and above versions throws errors, i.e: Kernel BPF verifier not accepted this
> existing BPF byte-code and system calls return error code "-7" as follows:
>
> rss_add_actions(): Failed to load BPF section l3_l4 (7): Argument list
> too long
>
>
> RCA: These errors started coming after from the Kernel-5.15 version, in
> which lots of new BPF verification restrictions were added for safe execution
> of byte-code on to the Kernel, due to which existing BPF program verification
> does not pass.
> Here are the major BPF verifier restrictions observed:
> 1) Need to use new BPF maps structure.
> 2) Kernel SKB data pointer access not allowed.
> 3) Undefined loops were not allowed(which are bounded by a variable value).
> 4) unreachable instructions(like: undefined array access).
>
> After addressing all these Kernel BPF verifier restrictions able to load the
> BPF byte-code onto the Kernel successfully.
>
> Note: This new BPF changes supports from Kernel:4.10 version.
>
> Bugzilla Id: 1329
>
> Signed-off-by: Madhuker Mythri
I tried this version on Debian testing which has:
kernel 6.5.0-5-amd64
clang 16.0.6
If build and run with the pre-compiled BPF then it will load the example flow
(see
https://urldefense.com/v3/__https://doc.dpdk.org/guides/nics/tap.html__;!!ACWV5N9M2RV99hQ!I6LFEsrAnlW2WMQlCy7Sxw-9MxJ_Qtchg-aZdal53Np6QtmC1wWsMdG_uT3zc57Yu7kvJANA-dQIirEUr-BWfaVXhGtd$
)
But if I recompile the bpf program by using make in the tap/bpf directory, then
the resulting bpf instructions will not make it past verifier.
With modified tap_bpf_api can get the log message as:
testpmd> flow create 0 priority 4 ingress pattern eth dst is
testpmd> 0a:0b:0c:0d:0e:0f / ipv4 / tcp / end actions rss queues 0 1 2
testpmd> 3 end / end
rss_add_actions(): Failed to load BPF section l3_l4 (13): func#0 @0
0: R1=ctx(off=0,imm=0) R10=fp0
0: (bf) r6 = r1 ; R1=ctx(off=0,imm=0)
R6_w=ctx(off=0,imm=0)
1: (18) r1 = 0x300; R1_w=768
3: (63) *(u32 *)(r10 -84) = r1; R1_w=768 R10=fp0 fp-88=
4: (bf) r2 = r10 ; R2_w=fp0 R10=fp0
5: (07) r2 += -84 ; R2_w=fp-84
6: (18) r1 = 0xfd ; R1_w=253
8: (85) call bpf_map_lookup_elem#1
R1 type=scalar expected=map_ptr
processed 7 insns (limit 100) max_states_per_insn 0 total_states 0
peak_states 0 mark_read 0
port_flow_complain(): Caught PMD error type 16 (specific action): cause:
0x7ffcef37e678, action not supported: Operation not supported
