feat(ber,power): Add the BER sweep and TR518 load; keep ce0 actually sending
Three new tools and one fix to an old one.
blanton_ber.sh - PRBS BER across every cabled port, init/start/report/
stop/clear, -u 0|1|all. A lane passes only below 1e-6 (equal to the
threshold fails). report ends with a per-unit summary carrying the worst
lane, so an all-PASS run still shows how much margin there was. init
checks every lane locked, because an unlocked lane still reports a BER
and it is a meaningless one. Script C runs it after the traffic report,
following SWB_UNIT0/SWB_UNIT1 like the traffic block does.
blanton_tr518.sh - the built-in packet test, used to load the rails
before measuring power. start settles 120s so the reading afterwards is
loaded rather than mid-ramp. stop sends port cd lb=none and nothing else,
exactly what the bench procedure does - l2 learning and test mode are
left as start set them, and stop says so instead of quietly leaving the
box in a state nobody asked about.
TH6_SWB{0,1}_power_readback.sh - 19 rails, V x I and total. These read
the two sensor tables ONCE. The previous version read them per rail: 38
invocations of a slow CLI at 38 different instants, so the "total" was a
sum of readings seconds apart - and under load the currents move (TH6_CORE
186 A -> 636 A). A rail missing from the table now warns on stderr instead
of silently contributing 0 W.
blanton_traffic_linespeed.sh - ce0 (VLAN 138) joins the test. It has no
loopback partner and a switch never sends a frame back out its ingress
port, so a tx burst was one round trip and then silence. init now sets an
ingress mirror of ce0 back to itself; the mirrored copy is not
egress-filtered, so packets keep lapping until stop turns the mirror off.
stop is what ends it - removing the VLAN membership does not, because the
copy never consulted the VLAN. TL_CE_MODE also drops to 'self': ps ce
shows one ce port per unit and a unit-0-only burst returned on unit 0.
Read as cross-unit, both links dying would still report PASS (0 == 0).
Fixes a parser bug worth remembering: bcmcmd returns CRLF, and the BER is
the last token on its line, so the CR stayed glued to the number. Every
lane read as NA, and printing the CR sent the cursor to column 1 so the
RESULT column overwrote the start of the row. One stray \r, two symptoms.
Not yet verified on hardware, all called out in the release notes: the
ce0 mirror (destport == srcport may be refused), the 120s settle (tr 518
may be time-limited), and 'test mode nr=no' on stop.
Script A -> V1.1.0.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014RetWKFZFG1ZHcitQAyhwM
This commit is contained in:
Claude Opus 5
@@ -9,7 +9,7 @@ DUT 端的取數能力由一組 **bash 工具庫** 提供（CB FPGA `pcimem` →
|
||||
|
||||
完整架構見 [ARCHITECTURE.md](ARCHITECTURE.md)。
|
||||
|
||||
**目前狀態（2026-08-27）**：Script A 為 **V1.0.10**（A 集中記錄 A/B/C 三支的變更）。涵蓋
|
||||
**目前狀態（2026-08-28）**：Script A 為 **V1.1.0**（A 集中記錄 A/B/C 三支的變更）。涵蓋
|
||||
PCIe AER / DDR EDAC / PMON+TPM / 電壓 margin（9 顆 LTC2980）/ SWB loopback 線速流量 /
|
||||
10G+1G 管理網路 ping / BMC DDR 與 USB / NVMe 健康，壓力層走平台的 `bgctl`。
|
||||
switch unit 數量由 `config.ttl` 的 `SWB_UNIT0` / `SWB_UNIT1` 宣告，風扇轉速由 `FAN_SPEED` 指定
|
||||
@@ -17,7 +17,10 @@ switch unit 數量由 `config.ttl` 的 `SWB_UNIT0` / `SWB_UNIT1` 宣告，風扇
|
||||
測試 job log 收在 DUT 的 `/host/hw-eval/jobs/`，Script C 結束時掛好 USB 再帶時戳複製到 `/mnt/usb/`。
|
||||
100G PRBS（`port_prbs_monitor.sh`）已隨包交付，但 **A/B/C 的呼叫仍註解掉**，還在 bring-up。
|
||||
Script B 的 soak 迴圈每輪印 `bgctl list` + `jobs`，並 `pause 60`（V1.0.10 之前完全沒有 pause）。
|
||||
最後一次發布是 **V1.0.9**；SWB 的 I2C 通道與位址已於 2026-08-19 上機驗證。
|
||||
V1.1.0 新增全機 SerDes BER 掃描（`blanton_ber.sh`，**Script C 已接**，依 `SWB_UNIT` 分支）、
|
||||
TR518 負載（`blanton_tr518.sh`，**TTL 尚未接**）、SWB 軌功耗（`TH6_SWB{0,1}_power_readback.sh`），
|
||||
線速流量也納入 100G `ce0`（VLAN 138，靠 ingress mirror 維持循環）。
|
||||
最後一次發布是 **V1.0.11**；SWB 的 I2C 通道與位址已於 2026-08-19 上機驗證。
|
||||
交付用 `./publish/publish.sh` 打包。
|
||||
|
||||
## 技術棧
|
||||
@@ -107,6 +110,23 @@ bash ~/Blanton_Script/port_prbs_monitor.sh report # PASS/FAIL 統計（沒
|
||||
bash ~/Blanton_Script/port_prbs_monitor.sh stop # 收尾會跑 prbsstat STOp + prbs clear
|
||||
# ⚠️ PRBS 打起來時 link 會顯示 down，這是正常的 —— 所以 start 刻意不印 port status
|
||||
|
||||
# ── 全機 SerDes BER（PRBS，期間 link 會顯示 DOWN）────
|
||||
./Blanton_Script/blanton_ber.sh init # prbs set p=3 + prbs get（檢查每條 lane 鎖定）
|
||||
./Blanton_Script/blanton_ber.sh start # prbsstat STArt Interval=30
|
||||
sleep 30 # 至少等一個 interval 再 report
|
||||
./Blanton_Script/blanton_ber.sh report # 每 lane 一列，BER < 1e-6 才 PASS，最後 summary
|
||||
./Blanton_Script/blanton_ber.sh stop # prbsstat STOp
|
||||
./Blanton_Script/blanton_ber.sh clear # prbs clear（會解除 PRBS，要重測得再 init）
|
||||
# -u 0 / -u 1 單邊；-c 1e-9 換門檻；-F 只列失敗；-f <log> 離線重解
|
||||
# ⚠️ 不能跟 blanton_traffic_linespeed / blanton_tr518 同時跑（都會接管同一批 port）
|
||||
|
||||
# ── TR518 負載 + 量功耗（TTL 尚未接）─────────────────
|
||||
./Blanton_Script/blanton_tr518.sh start # 兩 unit，跑完 settle 120s
|
||||
./Blanton_Script/TH6_SWB0_power_readback.sh # 19 條軌 V×I + TOTAL POWER
|
||||
./Blanton_Script/TH6_SWB1_power_readback.sh
|
||||
./Blanton_Script/blanton_tr518.sh stop # 只送 port cd lb=none
|
||||
# ⚠️ stop 不還原 l2 learn / test mode（TR_RESTORE_LEARN=1 / _TESTMODE=1 才會）
|
||||
|
||||
# ── USB 目標偵測 ────────────────────────────────────
|
||||
bash ~/Blanton_Script/usb_target.sh -o dev # /dev/sdX1
|
||||
bash ~/Blanton_Script/usb_target.sh -o mnt # 掛好並回傳掛載點
|
||||
@@ -235,6 +255,17 @@ Tera Term 連 COM port (115200-8-N-1)
|
||||
（版本改了、字串換了）就會永遠卡住且不報錯 —— 現場看起來像 Script A 當掉
|
||||
- ⚠️ **`show_dmesg.ttl` 結尾會 `dmesg -C` 清空 kernel ring buffer**。這是刻意的（讓 Script C 只看到
|
||||
soak 期間新產生的訊息），但代表事後在 DUT 上 `dmesg` 撈不到舊訊息 —— 內容只存在 master log 裡
|
||||
- ⚠️ **`bcmcmd` 的輸出是 CRLF**。`show c` 的計數器後面還跟著 delta，所以「砍掉第一個空白之後」
|
||||
順便把 CR 砍掉了；但 `prbsstat Ber` 的 BER 是**整行最後一個 token**，CR 直接黏在數字上 ——
|
||||
數字驗證會失敗（全變 NA），而且印出來時 CR 把游標拉回行首，後面的欄位會**覆蓋掉前面的欄位**。
|
||||
寫新的 awk 解析器時，第一件事就是 `{ gsub(/\r/, "") }`（`blanton_ber.sh` V1.0.1 的教訓）
|
||||
- ⚠️ **`bcmcmd` 的 exit code 兩邊都不能信**：BCM shell 拒絕指令時它回 **0**
|
||||
（`blanton_traffic_linespeed.sh` V0.3.1），而 `tr 518` 這種跑很久的指令會讓它 poll 逾時而回
|
||||
**非 0**，但指令其實已經在跑（`polling socket timeout: Success`）。可靠的作法是**找成功的證據**
|
||||
——例如 `tr 518` 印出 `TR 518 TEST PARAMETERS` 才算收下（`blanton_tr518.sh` V1.0.1）
|
||||
- ⚠️ **交換晶片不會把封包從進來的那個 port 送回去**（source-port knockout）。所以單一 port 的
|
||||
VLAN 打不出持續流量：`tx 100` 就是一趟來回然後結束。`ce0` 靠 **ingress mirror 打回自己**繞過
|
||||
這個限制（鏡像複本不經過 egress filter），**且不 stop 就不會自己停**
|
||||
- ⚠️ **DUT 的 awk 對 `%d` 會夾到 INT32**（2147483647）。線速流量的計數器是 ~10¹² 量級，
|
||||
所以 awk 裡輸出大數一律用 `%.0f`（double 域，精確到 2⁵³）。`blanton_traffic_linespeed.sh`
|
||||
V0.4.1 已修；寫新的 awk 報表時要記得同一件事
|
||||
|
||||
Reference in New Issue
Block a user