Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

scheduler: "penalise" some subflows by sending less than their cwnd #345

Open
Tracked by #350
matttbe opened this issue Feb 1, 2023 · 2 comments
Open
Tracked by #350
Labels
enhancement sched packets scheduler

Comments

@matttbe
Copy link
Member

matttbe commented Feb 1, 2023

Resources might be limited at MPTCP level (sending/receiving window). Also some terrible subflows can badly impact the performances of the MPTCP subflows.

MPTCP has a view of all the different subflows and it can tell which subflow is "bad" according to different criteria: high latency, losses, with bufferbloat, unstable, stale, etc.
The packet scheduler should then use the limited resources the best way and not just fill the cwnd of all subflows!

Such optimisation is in place in mptcp.org, see mptcp_rcv_buf_optimization(). In mptcp.org, the cwnd of the slow flows is halved max once per RTT but it looks like a hack: it is strange to modify this without telling the CC and the CC will potentially reset the modification quickly after.

Instead, the idea here would be to keep some states per subflow (similar to #342) to send a fraction of the cwnd. It might be needed to do more than just halving the cwnd. Probably keeping a shift and a multiplication is enough (for the moment): (cwnd >> x) * y, x and y being u8 values. (Maybe new hooks for new schedulers will be needed but that will be evaluated later, in a different ticket)

It is also important to reset the penalisation at some points. Would it be handled by the core (after each burst?) or by the scheduler (e.g. checking after each received ACK and once per RTT if this is still needed)?

The default scheduler should do that and new ones should be able to change the default behaviour.

@sferlin
Copy link

sferlin commented May 3, 2023

To the comment "The packet scheduler should then use the limited resources the best way and not just fill the cwnd of all subflows!"
This is the design of some schedulers, e.g., BLEST.

"the cwnd of the slow flows is halved max once per RTT but it looks like a hack: it is strange to modify this without telling the CC and the CC will potentially reset the modification quickly after."
This is the bevaviour of the penalisation&retransmission algorithm, correct? IMHO, CWND change is the job of the CC and not from an outer loop to interfere in the operation. That said, the scheduler has to make better predictions about the situation of each subflow and schedule data in the CWND space offered by the CC.

I opt to completely remove the penalisation&retransmission loop from the scheduler and CC operation altogether in MPTCP. This has been done while working on scheduler algorithms, e.g., BLEST. If this is not possible, I would suggest to limit its operation when minRTT is selected, as this was the default scheduler when P&R was designed. Other schedulers that came thereafter (whether default or not) had often not considered P&R operation in their loops.

@matttbe
Copy link
Member Author

matttbe commented Jun 2, 2023

the cwnd of the slow flows is halved max once per RTT but it looks like a hack: it is strange to modify this without telling the CC and the CC will potentially reset the modification quickly after.

This is the bevaviour of the penalisation&retransmission algorithm, correct?

@sferlin that's the behaviour of the out-of-tree kernel. We don't do that in the upstream kernel.

IMHO, CWND change is the job of the CC and not from an outer loop to interfere in the operation. That said, the scheduler has to make better predictions about the situation of each subflow and schedule data in the CWND space offered by the CC.

Yes, we agree on that! That's why we would prefer not to modify the CWND from the packet scheduler but use a part of it (so at least saving what part of the CWND we are using + maybe a timestamps)

I opt to completely remove the penalisation&retransmission loop from the scheduler and CC operation altogether in MPTCP. This has been done while working on scheduler algorithms, e.g., BLEST.

In the upstream kernel, we currently don't do that (compared to the out-of-tree kernel and the BLEST scheduler in this version also does that because it uses mptcp_next_segment() like the default scheduler and mptcp_rcv_buf_optimization() is called from there) and I think we need a way to limit the utilisation of one subflow. Currently in the upstream kernel with a BLEST-like implementation (best to ask Paolo for more details :-) ), we are impacted by subflows taking all resources: e.g. very high latency or losses, etc.

matttbe pushed a commit that referenced this issue Mar 27, 2024
A64_LDRSW() takes three registers: Xt, Xn, Xm as arguments and it loads
and sign extends the value at address Xn + Xm into register Xt.

Currently, the offset is being directly used in place of the tmp
register which has the offset already loaded by the last emitted
instruction.

This will cause JIT failures. The easiest way to reproduce this is to
test the following code through test_bpf module:

{
	"BPF_LDX_MEMSX | BPF_W",
	.u.insns_int = {
		BPF_LD_IMM64(R1, 0x00000000deadbeefULL),
		BPF_LD_IMM64(R2, 0xffffffffdeadbeefULL),
		BPF_STX_MEM(BPF_DW, R10, R1, -7),
		BPF_LDX_MEMSX(BPF_W, R0, R10, -7),
		BPF_JMP_REG(BPF_JNE, R0, R2, 1),
		BPF_ALU64_IMM(BPF_MOV, R0, 0),
		BPF_EXIT_INSN(),
	},
	INTERNAL,
	{ },
	{ { 0, 0 } },
	.stack_depth = 7,
},

We need to use the offset as -7 to trigger this code path, there could
be other valid ways to trigger this from proper BPF programs as well.

This code is rejected by the JIT because -7 is passed to A64_LDRSW() but
it expects a valid register (0 - 31).

 roott@pjy:~# modprobe test_bpf test_name="BPF_LDX_MEMSX | BPF_W"
 [11300.490371] test_bpf: test_bpf: set 'test_bpf' as the default test_suite.
 [11300.491750] test_bpf: #345 BPF_LDX_MEMSX | BPF_W
 [11300.493179] aarch64_insn_encode_register: unknown register encoding -7
 [11300.494133] aarch64_insn_encode_register: unknown register encoding -7
 [11300.495292] FAIL to select_runtime err=-524
 [11300.496804] test_bpf: Summary: 0 PASSED, 1 FAILED, [0/0 JIT'ed]
 modprobe: ERROR: could not insert 'test_bpf': Invalid argument

Applying this patch fixes the issue.

 root@pjy:~# modprobe test_bpf test_name="BPF_LDX_MEMSX | BPF_W"
 [  292.837436] test_bpf: test_bpf: set 'test_bpf' as the default test_suite.
 [  292.839416] test_bpf: #345 BPF_LDX_MEMSX | BPF_W jited:1 156 PASS
 [  292.844794] test_bpf: Summary: 1 PASSED, 0 FAILED, [1/1 JIT'ed]

Fixes: cc88f54 ("bpf, arm64: Support sign-extension load instructions")
Signed-off-by: Puranjay Mohan <puranjay12@gmail.com>
Message-ID: <20240312235917.103626-1-puranjay12@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
matttbe pushed a commit that referenced this issue May 20, 2024
Recent additions in BPF like cpu v4 instructions, test_bpf module
exhibits the following failures:

  test_bpf: #82 ALU_MOVSX | BPF_B jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #83 ALU_MOVSX | BPF_H jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #84 ALU64_MOVSX | BPF_B jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #85 ALU64_MOVSX | BPF_H jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #86 ALU64_MOVSX | BPF_W jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)

  test_bpf: #165 ALU_SDIV_X: -6 / 2 = -3 jited:1 ret 2147483645 != -3 (0x7ffffffd != 0xfffffffd)FAIL (1 times)
  test_bpf: #166 ALU_SDIV_K: -6 / 2 = -3 jited:1 ret 2147483645 != -3 (0x7ffffffd != 0xfffffffd)FAIL (1 times)

  test_bpf: #169 ALU_SMOD_X: -7 % 2 = -1 jited:1 ret 1 != -1 (0x1 != 0xffffffff)FAIL (1 times)
  test_bpf: #170 ALU_SMOD_K: -7 % 2 = -1 jited:1 ret 1 != -1 (0x1 != 0xffffffff)FAIL (1 times)

  test_bpf: #172 ALU64_SMOD_K: -7 % 2 = -1 jited:1 ret 1 != -1 (0x1 != 0xffffffff)FAIL (1 times)

  test_bpf: #313 BSWAP 16: 0x0123456789abcdef -> 0xefcd
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 301 PASS
  test_bpf: #314 BSWAP 32: 0x0123456789abcdef -> 0xefcdab89
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 555 PASS
  test_bpf: #315 BSWAP 64: 0x0123456789abcdef -> 0x67452301
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 268 PASS
  test_bpf: #316 BSWAP 64: 0x0123456789abcdef >> 32 -> 0xefcdab89
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 269 PASS
  test_bpf: #317 BSWAP 16: 0xfedcba9876543210 -> 0x1032
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 460 PASS
  test_bpf: #318 BSWAP 32: 0xfedcba9876543210 -> 0x10325476
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 320 PASS
  test_bpf: #319 BSWAP 64: 0xfedcba9876543210 -> 0x98badcfe
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 222 PASS
  test_bpf: #320 BSWAP 64: 0xfedcba9876543210 >> 32 -> 0x10325476
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 273 PASS

  test_bpf: #344 BPF_LDX_MEMSX | BPF_B
  eBPF filter opcode 0091 (@5) unsupported
  jited:0 432 PASS
  test_bpf: #345 BPF_LDX_MEMSX | BPF_H
  eBPF filter opcode 0089 (@5) unsupported
  jited:0 381 PASS
  test_bpf: #346 BPF_LDX_MEMSX | BPF_W
  eBPF filter opcode 0081 (@5) unsupported
  jited:0 505 PASS

  test_bpf: #490 JMP32_JA: Unconditional jump: if (true) return 1
  eBPF filter opcode 0006 (@1) unsupported
  jited:0 261 PASS

  test_bpf: Summary: 1040 PASSED, 10 FAILED, [924/1038 JIT'ed]

Fix them by adding missing processing.

Fixes: daabb2b ("bpf/tests: add tests for cpuv4 instructions")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/91de862dda99d170697eb79ffb478678af7e0b27.1709652689.git.christophe.leroy@csgroup.eu
matttbe pushed a commit that referenced this issue May 28, 2024
Add a test case to assert that the skb->pkt_type which was set from the BPF
program is retained from the netkit xmit side to the peer's device at tcx
ingress location.

  # ./vmtest.sh -- ./test_progs -t netkit
  [...]
  ./test_progs -t netkit
  [    1.140780] bpf_testmod: loading out-of-tree module taints kernel.
  [    1.141127] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel
  [    1.284601] tsc: Refined TSC clocksource calibration: 3408.006 MHz
  [    1.286672] clocksource: tsc: mask: 0xffffffffffffffff max_cycles: 0x311fd9b189d, max_idle_ns: 440795225691 ns
  [    1.290384] clocksource: Switched to clocksource tsc
  #345     tc_netkit_basic:OK
  #346     tc_netkit_device:OK
  #347     tc_netkit_multi_links:OK
  #348     tc_netkit_multi_opts:OK
  #349     tc_netkit_neigh_links:OK
  #350     tc_netkit_pkt_type:OK
  Summary: 6/0 PASSED, 0 SKIPPED, 0 FAILED

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/r/20240524163619.26001-4-daniel@iogearbox.net
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement sched packets scheduler
Projects
Status: Needs triage
MPTCP Future
  
Needs triage
Development

No branches or pull requests

2 participants