Commit Graph

6165 Commits

Author SHA1 Message Date
whitequark a8615cdbcf firmware: globally enable LTO.
This used to crash with earlier rustc versions, but doesn't anymore,
and gives significant speedup (e.g. 2x on test_dma_record_time).
2018-08-12 19:11:48 +00:00
whitequark 5da45f9486 firmware: migrate to Rust 1.28.0.
This also updates / is a prerequisite for updating smoltcp.

Rationale for changes made:
  * compiler_builtins is now shipped in the rust prefix.
  * rustc's libpanic_unwind no longer works for us because it
    has a hard dependency on Box (and it's a horrible hack);
    fortunately, we only ever needed a personality function
    from it.
  * panic and oom handlers are now set in a completely different
    way.
  * allocators are quite different (and finally stable).
  * NLL caused internal compiler errors in runtime, so code using
    NLL was rewritten to not rely on it and it was turned off.
2018-08-12 18:40:08 +00:00
Sebastien Bourdeauducq 738d2c6bcb hmc7043: REFSYNCIN → RFSYNCIN 2018-08-11 12:07:17 +08:00
Sebastien Bourdeauducq bc3e715a8f examples: fix kasli_tester 2018-08-11 10:51:42 +08:00
whitequark fab6e5cdff compiler: skip functional values in attribute writeback.
Fixes #1088.
2018-08-10 12:02:49 +00:00
Sebastien Bourdeauducq 052e400f12 test: skip test_dma_playback_time on Kasli (#946) 2018-08-09 18:08:21 +08:00
Sebastien Bourdeauducq 957645a7e7 examples: move kasli tester out of kasli_basic 2018-08-09 18:07:44 +08:00
Sebastien Bourdeauducq bbc98410e4 test: dds → ad9914dds
Prevent confusion with Urukul.
2018-08-09 16:55:09 +08:00
Sebastien Bourdeauducq bf78e0c7d2 test: fix handling of missing devices 2018-08-09 16:51:12 +08:00
Robert Jördens a061ba2505 grabber/kasli_basic: add grabber test
close #1121
2018-08-08 12:43:44 +02:00
Robert Jördens f7678cc24a grabber: refactor state machine 2018-08-07 18:07:46 +02:00
Robert Jördens 6cd2432e30 grabber: log all resolution changes
close #1120
2018-08-07 16:21:21 +02:00
Robert Jördens 99a15ca0c6 grabber: rationalize derived traits 2018-08-07 16:21:21 +02:00
Sebastien Bourdeauducq 49f7a1610f sayma: use GTP_CLK1 only for all variants (#1080) 2018-08-07 20:53:14 +08:00
Sebastien Bourdeauducq e2a49ce368 drtio: support external IBUFDS_GTE3 2018-08-07 20:52:45 +08:00
Sebastien Bourdeauducq 8b8e1844f0 kasli_sawgmaster: roughly match Urukul and Sayma amplitudes 2018-08-07 20:07:21 +08:00
Sebastien Bourdeauducq 9ce6233926 kasli: fix SYSU TTL directions 2018-08-07 19:29:28 +08:00
Sebastien Bourdeauducq 8aa88cfe70 kasli_sawgmaster: add Urukul-Sayma example 2018-08-07 19:29:28 +08:00
Robert Jördens 474bc7b65b browser: handle windows file urls for feeding h5py
close #1014
2018-08-07 12:57:01 +02:00
whitequark 93af5d2a03 compiler: handle async RPC as last statement in try block.
Fixes #1107.
2018-08-07 07:06:53 +00:00
whitequark 7bd7b6592a rpc_proto: serialize keywords correctly.
Fixes #1109.
2018-08-07 06:47:09 +00:00
whitequark 259f1576c3 Fix tests after a74958f0. 2018-08-07 06:06:49 +00:00
whitequark a74958f01f ksupport: raise RuntimeError on reraise with no inflight exception.
Fixes #1123.
2018-08-07 05:53:13 +00:00
Sebastien Bourdeauducq 2008d02f4d runtime: use different default IP and MAC for different kinds of boards
This helps reduce conflicts when having many boards on a development network.
2018-08-07 10:30:50 +08:00
Sebastien Bourdeauducq bbe36b94f7 ad9154: enable sync in init 2018-08-06 19:02:27 +08:00
Sebastien Bourdeauducq 7f0b2ff594 jesd204sync: work around HMC7043 poor behavior with combined delays
The HMC7043 outputs poorly controlled signals when adjusting
two delays at once. This commit puts the DAC in one-shot SYSREF mode,
and only triggers synchronizations when SYSREF is stable.
2018-08-06 17:43:17 +08:00
Sebastien Bourdeauducq f32f0126e2 Revert "ad9154: use continuous sync mode"
The HMC7043 is not really glitchless.

This reverts commit bd968211de.
2018-08-06 16:59:53 +08:00
Sebastien Bourdeauducq 65f198bdee kasli: use tester EEMs for DRTIO, add Urukul-Sayma sync example 2018-08-06 16:53:13 +08:00
Sebastien Bourdeauducq bd968211de ad9154: use continuous sync mode 2018-08-06 00:27:10 +08:00
Sebastien Bourdeauducq b023865b42 sayma: instantiate dummy IBUFDS_GTE3 on unused but driven Si5324 clock pins
Solve same problem as e83ee3a0 but channels cannot be independently disabled.
2018-08-05 23:02:41 +08:00
Sebastien Bourdeauducq e83ee3a07a hmc7043: disable GTP_CLK1 when not in use
Termination and biasing are not active at the FPGA when IBUFDS_GTE3 is
not instantiated, and driving a clock then leads to overvoltage.
2018-08-03 10:03:52 +08:00
Chris Ballance 6fc8439399 tweak moninj to allow old dashboard with new firmware 2018-08-02 19:34:14 +08:00
Chris Ballance 04cbc3237b test_moninj: test injection monitoring 2018-08-02 19:34:14 +08:00
Chris Ballance 47740c8930 share moninj injection state between dashboards
Previously if one dashboard overrode a channel this was not visible on
any other dashboard - the channel appeared to operate normally.
2018-08-02 19:34:14 +08:00
Robert Jördens 7d6a1b528d ad9912: add ftw_to_frequency 2018-08-02 11:19:12 +00:00
Robert Jördens e518a1f1d0 ad53xx: also increase slack after control readback 2018-08-02 11:19:12 +00:00
whitequark 5871d13da8 firmware: actually compact in config::compact().
Fixes #1116.
2018-08-01 16:27:48 +00:00
David Nadlinger 829fca6112 pyon: Correctly deserialize bare NaNs
This also fixes (non-numpy) lists containing NaNs.

Previously, accidentally storing a NaN in a dataset would
bring down large parts of the system.
2018-07-30 11:08:56 +01:00
David Nadlinger 08ee91beb2 ad9910: Clarify chip_select range [nfc]
`assert 3 <= chip_select <= 7` is rather opaque without looking
at the CPLD source code otherwise.
2018-07-30 11:08:17 +01:00
David Nadlinger 6b89106578 ad53xx: Avoid sporadic RTIOUnderflow in init()
I observed sporadic RTIO underflows on Kasli before, by ~2.5 µs,
so 5 µs extra slack should be plenty. No underflows since.
2018-07-28 23:45:48 +01:00
Sebastien Bourdeauducq c00eb181fc conda: bump jesd204b 2018-07-28 13:17:44 +08:00
Sebastien Bourdeauducq e4d48a78eb drtio: wait for remote to ack TSC synchronization
Sayma takes a long time after TSC sync to align SYSREF, and this caused two issues:
1. Aux packets getting lost and causing error reports
2. DRTIO links reported up and kernels proceeding despite the DACs not being properly synced.
2018-07-26 20:28:17 +08:00
Sebastien Bourdeauducq 83de8b2ba2 drtio: add ping timeout during link init 2018-07-26 20:27:53 +08:00
Sebastien Bourdeauducq 446f791180 firmware: simplify SYSREF DRTIO alignment 2018-07-26 19:37:59 +08:00
Sebastien Bourdeauducq f8c17528e7 satman: use new SYSREF code 2018-07-26 16:26:57 +08:00
Sebastien Bourdeauducq 32c95ac034 sayma: automated DAC SYSREF phase calibration 2018-07-26 16:23:55 +08:00
Sebastien Bourdeauducq dbcf2fe9b4 firmware: remove 'chip found' messages on Sayma 2018-07-26 16:07:37 +08:00
Sebastien Bourdeauducq d523d03f71 sayma: automated FPGA SYSREF phase offset calibration 2018-07-26 14:53:28 +08:00
Sebastien Bourdeauducq 0a9d3638ee config: add write_int 2018-07-26 14:49:32 +08:00
Sebastien Bourdeauducq 19c51c644e grabber: cleanup GRABBER_STATE 2018-07-24 19:08:51 +08:00