riscv/rocket-chip
1
0
Fork 0
Code Releases Activity
3,813 Commits 1 Branch 0 Tags 13 MiB
37a3c22639
Commit Graph

958 Commits

Author SHA1 Message Date
Howard Mao
fa8844d5c3 properly use rocket MT_ constants in regression tests 2016-10-28 18:26:34 -07:00
Andrew Waterman
e45b41b4b6 Don't rely on SeqMem output after read-enable is low 2016-10-27 23:44:10 -07:00
Howard Mao
900a7bbcf1 add PutAtomic support to width adapter 2016-10-26 09:58:26 -07:00
Jacob Chang
fc5eb7cc64 Fixed AsyncFifo with reset messaging 2016-10-25 16:45:08 -07:00
Megan Wachs
fd2d48acda lazy_module: If the user actually specifies a name, just use it without appending module name. 2016-10-25 15:58:09 -07:00
Wesley W. Terpstra
a807c922d0 diplomacy: take names from the outermost common node 2016-10-25 15:58:09 -07:00
Wesley W. Terpstra
fee67c4abf diplomacy: add methods to find {out,in}ner-most common node 2016-10-25 15:58:09 -07:00
Wesley W. Terpstra
67ab27f5a5 diplomacy: guess the LazyModule name from the containing class 2016-10-25 15:58:09 -07:00
Wesley W. Terpstra
4d50733548 tilelink2 ToAXI4: use helper method for a_last (#418) 2016-10-25 10:16:42 -07:00
Wesley W. Terpstra
7dc97674d6 rocketchip: include an socBus between l1tol2 and periphery (#415) 
Sometimes we have high performance devices that go inbetween.
 2016-10-24 23:56:09 -07:00
Wesley W. Terpstra
a5ac106bb8 axi4 ToTL: fix decode error arbitration (#417) 
When selecting between error generation on R and real data on R,
correctly calculate the R backpressure.

This bug manifests when a valid request is immediately followed by
an invalid request, wedging the R channel.
 2016-10-24 22:15:19 -07:00
Wesley W. Terpstra
4c815f7958 tilelink2 Parameters: fix {contains,supports}Safe (#416) 
When there is only one manager, you still want to know if the address
was wrong on the link to that manager!
 2016-10-24 20:37:04 -07:00
Wesley W. Terpstra
8bfd6bcd4d axi4: ensure we accept AR before reporting R (#411) 2016-10-21 21:02:05 -07:00
Colin Schmidt
85f3788ab5 initialize s2_hit to solve #401 2016-10-21 14:53:55 -07:00
Wesley W. Terpstra
7c334e3c34 axi4 ToTL: shorter critical path on Q.bits if errors go first 2016-10-17 01:00:49 -07:00
Wesley W. Terpstra
73010c79a3 axi4 ToTL: handle bad AXI addresses 2016-10-17 00:12:26 -07:00
Wesley W. Terpstra
501d6d689f axi4: Test ToTL 2016-10-16 22:04:06 -07:00
Wesley W. Terpstra
5a1da63b5a axi4: prototype ToTL adapter 2016-10-16 22:04:01 -07:00
Wesley W. Terpstra
72e5a97d40 tilelink2: factor out the OH1ToOH function 2016-10-16 22:04:01 -07:00
Wesley W. Terpstra
d09f43c32f axi4 Bundles: add a size calculation helper 
The old version was wrong.
Inverting before the << has a different width.
This means you end up with high bits set.
 2016-10-16 22:04:01 -07:00
Wesley W. Terpstra
20288729b9 tilelink2 Isolation: cross the valid signals as well 
Refactor the code to be less copy-pasty
 2016-10-14 18:28:36 -07:00
Wesley W. Terpstra
680a944f07 regmapper RegisterCrossing: safe AsyncQueues are overkill here 2016-10-14 18:28:31 -07:00
Wesley W. Terpstra
ac0bb841da AsyncQueue: cope with far reset propagation delay 2016-10-14 18:05:35 -07:00
Wesley W. Terpstra
8f3c2ddfc3 tilelink2 Crossing: these asserts should be done by the AsyncQueue 2016-10-14 16:54:09 -07:00
Wesley W. Terpstra
a82cfb8306 tilelink2: replace addr_hi with address (#397) 
When faced with ambiguous routing of wmask=0, we decided to include
all the address bits. Hopefully in most cases the low bits will be
optimized away anyway.
 2016-10-14 14:09:39 -07:00
Wesley W. Terpstra
4e40f9bb59 tilelink2 Nodes: appease the PC police 2016-10-13 17:02:18 -07:00
Wesley W. Terpstra
54b73aef57 tilelink2: WidthWidget and Fragmenter no longer erase latency 2016-10-13 17:02:18 -07:00
Wesley W. Terpstra
200cf3dd13 tilelink2 Nodes: include some options to test for conformance 2016-10-13 17:02:18 -07:00
Wesley W. Terpstra
5d5b5a66f4 tilelink2 RAMModel: fix a write-bad-data bug 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
e5a1483358 tilelink2 Fragmenter: eliminate most of the registers on A 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
99c7003d11 tilelink2: allow preemption of Fragmenter and WidthWidget 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
b42cfdc9dd tilelink2 Arbiter: there is only one winner 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
b6e9b0c558 tilelink2 Arbiter: allow preemption of first beat 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
0aebf9e341 tilelink2 ToAXI4: no arbitration path register needed 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
0e897b905f tilelink2 RegisterRouter: data path register is no longer required 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
c4eadd3ab3 tilelink2 Monitor: enforce stricter transaction ordering 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
d8a1163131 tilelink2 Monitor: don't enforce Irrevocable any more 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
405f66da32 tilelink2 WidthWidget: cope with Decoupled inputs 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
e2e72ac979 tilelink2 Fragmenter: cope with Decoupled input 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
023c6402e9 tilelink2: switch to DecoupledIO syntax 2016-10-13 17:02:17 -07:00
Wesley W. Terpstra
4c1c52486b axi4 Fragmenter: handle more inflight AXI requests than we have space 2016-10-13 15:52:32 -07:00
Wesley W. Terpstra
8005266131 axi4 Fragmenter: refine sideband FSM for case of last fragment 2016-10-13 15:52:32 -07:00
Wesley W. Terpstra
19064e602b axi4 Fragmenter: align all output accesses 
We promised the output is aligned. Make good on that!
 2016-10-13 15:52:27 -07:00
Wesley W. Terpstra
84be93f9f3 axi4 Fragmenter: confirm correct handling of last 2016-10-13 14:01:23 -07:00
Wesley W. Terpstra
1c79a23a8b axi4 Fragmenter: initialize error response to 0 2016-10-13 13:46:24 -07:00
Wesley W. Terpstra
958af132ba axi4 Fragmenter: optimize dynamic slave lookup 2016-10-12 17:29:38 -07:00
Wesley W. Terpstra
11169d155c axi4: add a Buffer to put between nodes 2016-10-12 17:08:52 -07:00
Wesley W. Terpstra
a9a3f7dd4e tilelink2 RAMModel: include name of test in output 2016-10-12 17:08:52 -07:00
Wesley W. Terpstra
345eefd81b axi4: include unit tests 2016-10-12 17:08:52 -07:00
Wesley W. Terpstra
a6c6d99848 axi4: prototype Fragmenter 2016-10-12 17:08:49 -07:00
Wesley W. Terpstra
c918aa6d89 axi4: name AdapterNode parameters properly 2016-10-12 17:02:02 -07:00
Wesley W. Terpstra
a423f97844 axi4: parameterized AXI master constraint for aligned access 2016-10-12 17:02:02 -07:00
Wesley W. Terpstra
673cf1fdb5 tilelink2 ToAXI4: must create irrevocable D for now 2016-10-12 17:02:01 -07:00
Wesley W. Terpstra
8e92ac32b7 tilelink2 ToAXI4: we need a Queue on B to guarantee deadlock freedom 2016-10-12 17:02:01 -07:00
Wesley W. Terpstra
38b6c1c820 tilelink2 axi4: RegisterRouter can cut ready dependency 2016-10-12 17:02:01 -07:00
Wesley W. Terpstra
dc26736f32 axi4 tilelink2: include minAlignment and maxAddress in slaves 2016-10-12 17:02:01 -07:00
Wesley W. Terpstra
538437384a tilelink2 Fragmenter: combine AccessAck errors 2016-10-12 17:02:01 -07:00
Wesley W. Terpstra
4caa543ad7 tilelink2: Fragmenter should not cut Acquire parameters 
The correct response to misuse is to fail a requirement check.
Pretending that things are not caches could lead to inconsistency.
 2016-10-11 22:38:03 -07:00
Wesley W. Terpstra
6336f94fa2 tilelink2: only caches can support B requests 2016-10-11 22:38:02 -07:00
Wesley W. Terpstra
4a975ca380 tilelink2: add a rightOR to go with our leftOR 2016-10-11 22:38:02 -07:00
Wesley W. Terpstra
b2a5d18e37 diplomacy: simplify address range fragmentation 2016-10-11 22:36:21 -07:00
Wesley W. Terpstra
b0e33f4a39 tilelink2: use TLArbiter in HintHandler 2016-10-10 13:15:28 -07:00
Wesley W. Terpstra
683a2e6785 tilelink2: refactor firstlast helper method 2016-10-10 13:15:28 -07:00
Wesley W. Terpstra
a404cd2abf tilelink2: use NodeHandle to restore Crossing.node API 2016-10-10 13:15:28 -07:00
Wesley W. Terpstra
876609eb0e diplomacy: add NodeHandles to support abstraction 2016-10-10 13:15:25 -07:00
Wesley W. Terpstra
97af07eb3e tilelink2: clarify use of Isolation 2016-10-10 13:13:32 -07:00
Wesley W. Terpstra
76388117bb regmapper: detect improper reset sequencing in RegisterCrossing 2016-10-10 13:13:32 -07:00
Wesley W. Terpstra
b5f5ef69c1 regmapper: eliminate race condition in RegisterCrossing bypass 2016-10-10 13:13:32 -07:00
Wesley W. Terpstra
f250426728 tilelink2: blow up if the channels carry data when they should not 2016-10-10 13:13:32 -07:00
Wesley W. Terpstra
1b09f1360d AsyncQueue: adjust register names to match vals 2016-10-10 13:13:32 -07:00
Wesley W. Terpstra
e7f8a7e9ea AsyncQueue: make it clear that the SyncChain is not Gray specific 2016-10-10 13:13:32 -07:00
Wesley W. Terpstra
52b8121e68 Apply "async_queue: Give names to all the registers which show up in the queue (#390)" 
Adjusted to include names for the new registers.
Changes to RegisterCrossing were discarded.
 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
ffb734ac0e AsyncQueue: disambiguiate the reset_n signal names 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
5ee53c61d6 util: clarify an AsyncQueue corner-case 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
609fd97a71 util: AsyncQueue detect power-down/reset of non-empty queue 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
75bb94017b util: resynchronize AsyncQueue counters when far side resets 
If the other clock domain is much faster than ours, it's reset
might be shorter than a single cycle in our domain. In that case,
we need to catch the reset and extend it.
 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
5e2609bdd2 AsyncQueueSource: don't feed reset into normal logic! 
There is no need to block writes to mem during reset.
The Queue must be empty anyway.
 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
2f6985efd3 crossings: use flip not flip() 
This seems to be the more common API
 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
6d6aa3eb13 tilelink2: Isolation must also connect reset_n 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
cb7b16f1a9 util: exchange resets between AsyncQueue source and sink 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
8c7d469a95 Revert "async_queue: Give names to all the registers which show up in the queue (#390)" 
This reverts commit a84a961a39.

The changes to RegisterCrossing.scala were unneeded after application of this branch.
The name changes made to the AsyncQueue.scala are reapplied at the end of this branch.
 2016-10-10 13:13:31 -07:00
Wesley W. Terpstra
adf5f1807b tilelink2: ToAXI4 bridge added 2016-10-10 11:21:50 -07:00
Wesley W. Terpstra
e856cbe3a6 axi4: SRAM for testing 2016-10-10 11:21:50 -07:00
Wesley W. Terpstra
abb02aa6f4 axi4: add a RegisterRouter for generic devices 2016-10-10 11:21:50 -07:00
Wesley W. Terpstra
2f7081aeaf tilelink2: make mask generation reusable 2016-10-10 11:21:50 -07:00
Wesley W. Terpstra
b29d34038e axi4: diplomacy capable AXI4 2016-10-10 11:21:50 -07:00
Wesley W. Terpstra
dcb9383568 PositionalMultiQueue: work around vcs Lint report 
Lint-[PCTIO-L] Ports coerced to inout
rocket-chip/vsim/generated-src/unittest.UncoreUnitTestConfig.v, 127524
"io_deq_0_valid"
  Port "io_deq_0_valid" declared as output in module "PositionalMultiQueue_16"
  may need to be inout. Coercing to inout.
 2016-10-10 11:21:49 -07:00
Wesley W. Terpstra
5d905a5310 PositionalMultiQueue: shared storage FIFO 1-push n-pop 2016-10-10 11:21:49 -07:00
mwachs5
3a1d8fe482 debug: use a different form of the crossing which doesn't create an AsyncScope (#394) 2016-10-09 20:33:18 -07:00
mwachs5
b5d4b72313 register_crossing: Remove the need for AsyncScope by specifying the master clock and reset. (#393) 2016-10-09 15:51:23 -07:00
Henry Cook
1e69a2dc1c [tilelink2] allow TL monitors to be globally enabled or disabled (#392) 2016-10-09 12:34:10 -07:00
Andrew Waterman
53360f4c2c Disable U-mode by default unless S-mode is present 2016-10-08 21:29:40 -07:00
Andrew Waterman
7f429e8799 Simplify AsyncResetReg 
No need for AsyncSetReg, as AsyncResetReg can be used exclusively with
inverted inputs.
 2016-10-08 21:29:40 -07:00
mwachs5
a84a961a39 async_queue: Give names to all the registers which show up in the queue (#390) 
This is to aid debugging but even more so for backend constraint writers, who generally
need predictable names for registers to set false paths, etc.
 2016-10-08 17:50:50 -07:00
Andrew Waterman
4fd03ffdf1 Fix PopCountAtLeast, un-breaking BTB 2016-10-07 21:20:40 -07:00
Wesley W. Terpstra
e5ac0f717f tilelink2: split isolation gates by direction 2016-10-07 12:03:43 -07:00
Albert Ou
ad618fd55d plic: Fix bit extraction 2016-10-06 18:05:03 -07:00
Andrew Waterman
b1c777c7a2 Fix PLIC enable bit access for #ints >= tlDataBits 2016-10-06 16:21:14 -07:00
Andrew Waterman
c22438b822 Fix an overly strict D$ assertion 2016-10-06 15:52:46 -07:00
Jacob Chang
fe641c14a1 tilelink2: Add support for different noise generator in fuzzer (#386) 2016-10-06 13:20:13 -07:00
Andrew Waterman
5980dc160f Don't allow multiple entries for same PC in BTB 
Necessary for RVC forward-progress guarantee.
 2016-10-06 11:30:45 -07:00
Andrew Waterman
eddf1679f5 Use <> instead of := for bi-directional connections 2016-10-04 22:29:39 -07:00
Andrew Waterman
67593fdf2d Explicitly zap some S-mode CSRs when not using S-mode 2016-10-04 22:29:39 -07:00
Andrew Waterman
968851f7e3 Default to configurable priorities 
up-to-7 levels is kind of arbitrary, but I'm unwilling to introduce
a new Parameter at the moment.
 2016-10-04 22:29:39 -07:00
mwachs5
e952f8f222 asyncqueue: Fix typo in the Async Queue (#381) 
* asyncqueue: Fix typo in the Async Queue that would cause the sync depth to be one less than expected.

* asyncqueue: Typo in the typo fix
 2016-10-04 21:02:06 -07:00
Andrew Waterman
064c9ebdc6 Don't report I$ fetch faults on TLB misses! 2016-10-04 14:37:25 -07:00
Andrew Waterman
516481b68b Improve back-to-back integer multiplication performance 
More exact hazard checking in the decode stage avoids a pipeline flush.
 2016-10-04 14:37:25 -07:00
Andrew Waterman
7b69f1f261 Don't enter D$ flush state machine if grant outstanding 2016-10-04 14:37:25 -07:00
Andrew Waterman
28beb33943 Make any intervening load/store/fence fail an LR/SC sequence 
This catches LR/SC misuses more quickly.
 2016-10-04 14:37:25 -07:00
Yunsup Lee
62954d543e correctly initialize the active flag 2016-10-03 17:56:30 -07:00
Wesley W. Terpstra
6ec2e7c5bd tilelink2: Legacy should preserve the access size (#378) 
* tilelink2: Legacy should preserve the access size
* Legacy: extract missing size information for TL1 Puts
 2016-10-03 17:25:31 -07:00
Wesley W. Terpstra
f05298d9bc tilelink2: move general-purpose code out of tilelink2 package 2016-10-03 16:22:28 -07:00
Wesley W. Terpstra
c85e42a303 tilelink2: Nodes should accept full PortParameters 
We need this for terminal clients/managers that bridge multiple
non-TL2 devices.
 2016-10-03 16:09:49 -07:00
Wesley W. Terpstra
f2ca2178bf graphML: CTO's like colour 2016-10-03 15:05:45 -07:00
Wesley W. Terpstra
fe0875b084 LazyModule: output final verilog Module name 2016-10-03 15:05:45 -07:00
Wesley W. Terpstra
0a4ef66894 BaseTop: record top module; more general than GraphML 2016-10-03 15:05:45 -07:00
Wesley W. Terpstra
52c1a053ff tilelink2 RegisterRouter: test fully Decoupled behaviour 2016-10-02 02:24:02 -07:00
Wesley W. Terpstra
422e6357a4 tilelink2 RegisterCrossing: Queues go from RV to Irrevocable 
AsyncQueue is still a Queue.
 2016-10-02 02:24:02 -07:00
Wesley W. Terpstra
02f89fb530 RegMapper: clarify interface is DecoupledIO 2016-10-02 02:24:02 -07:00
Wesley W. Terpstra
8a268268ad tilelink2 RegField: clarify restrictions on functions 
RegMapper is fundamentaly DecoupledIO.
Let the user take advantage of this.
Clarify that rules on data handling.
 2016-10-02 02:24:02 -07:00
Wesley W. Terpstra
bff0ffa428 tilelink2 RegisterRouter: fix output data glitches 
If a device changes a register while it's being read but not yet accepted,
this an lead to 'data' changing while 'valid' is high. A violation. The
problem is that RegMapper is fundamentally DecoupledIO. So fix it with a
Queue.
 2016-10-02 02:24:02 -07:00
Andrew Waterman
e0188f8aa4 Don't implicitly fence on CSR instructions 
CSRs that have an effect on I/O should use an explicit FENCE.
 2016-10-01 19:44:10 -07:00
Andrew Waterman
b772edcb1b Allow hit-under-MMIO and multiple MMIOs in blocking D$ 
The latter feature is by default disabled, since there aren't enough
ID bits.
 2016-10-01 19:44:05 -07:00
Megan Wachs
28eba9b5ac clint/plic: Move the default addresses 2016-10-01 15:46:55 -07:00
mwachs5
9a381e88d1 Suggest sane names for common objects (#369) 
* Suggest sane names for common objects frequently instantiated with factory methods

* Suggest names for common primitives using more Scala-esque Options
 2016-09-30 16:19:25 -07:00
Wesley W. Terpstra
0ebab0976a tilelink2 Isolation: add enable signal (#368) 2016-09-30 04:54:40 -07:00
Wesley W. Terpstra
d3547a6193 tilelink2: Isolation gate insertion module 2016-09-30 01:50:33 -07:00
Wesley W. Terpstra
9b0654be52 tilelink2 Crossing: helpful constructor objects 2016-09-30 01:48:47 -07:00
Wesley W. Terpstra
80f7bb49e3 tilelink2: helper objects operate on OutwardNodes 2016-09-30 01:39:35 -07:00
Howard Mao
4b86802b1a change the configuration interface of SlowIO 2016-09-29 22:16:53 -07:00
Wesley W. Terpstra
6d8c965f04 tilelink2 Crossing: cut the crossing between clock domains 2016-09-29 17:35:10 -07:00
Wesley W. Terpstra
20f42a8762 tilelink2: reuse the halves of the AsyncQueue 2016-09-29 17:35:08 -07:00
Wesley W. Terpstra
8e4c1e567c tilelink2: add types for a TL clockless interface 2016-09-29 17:33:11 -07:00
Wesley W. Terpstra
02ce8c2ca4 tilelink2 Nodes: rename RootNode => BaseNode 2016-09-29 17:33:11 -07:00
Wesley W. Terpstra
754fcf9831 tilelink2: rename BaseNode to SimpleNode 2016-09-29 17:33:11 -07:00
Wesley W. Terpstra
cfdb8ca797 tilelink2 LazyModule: remove obsolete connect method 2016-09-29 17:33:11 -07:00
Wesley W. Terpstra
f2e438833c tilelink2 Nodes: generalize a node into inner and outer halves 
This lets us create nodes which transform from one bus to another.
 2016-09-29 17:33:11 -07:00
Andrew Waterman
2bdf8c2be7 Merge branch 'master' into move-to-util 2016-09-29 14:42:11 -07:00
Howard Mao
ab3219cf6e don't use Scala to Chisel implicit conversions outside of rocket 2016-09-29 14:35:42 -07:00
Howard Mao
9910c69c67 Move a bunch more things into util package 
A lot of utility code was just being imported willy-nilly from one
package to another. This moves the common code into util to make things
more sensible. The code moved were

 * The AsyncQueue and AsyncDecoupledCrossing from junctions.
 * All of the code in rocket's util.scala
 * The BlackBox asynchronous reset registers from uncore.tilelink2
 * The implicit definitions from uncore.util
 2016-09-29 14:23:42 -07:00
Andrew Waterman
e928b741ce Default mtvec=0, not None 
Setting it to None was a mistake.  It makes it far harder to
diagnose boot bugs, as you end up fetching from random addreses
after trapping.
 2016-09-29 13:52:41 -07:00
Megan Wachs
45bd63fcc6 jtag: Prevent Debug RAM accesses from wrapping around, and bring the DTM closer to the Debug Spec 2016-09-29 13:49:14 -07:00
Megan Wachs
449d689a4e jtag: Connect the JTAG DTM side of the synchronizer! 2016-09-29 13:48:55 -07:00
Yunsup Lee
0924f8adb0 print out assigned inerrupt ranges 2016-09-29 11:59:32 -07:00
Yunsup Lee
4c3e8ec1b4 assign interrupt ranges deterministically 2016-09-29 11:59:32 -07:00
Henry Cook
7bca99a27a [tilelink2] Add unit test configs to regression 2016-09-28 18:02:04 -07:00
Henry Cook
32f3f94882 [tilelink2] Fix zero-width wires in RAMModel. 2016-09-28 18:02:04 -07:00
Henry Cook
69e121260e [tilelink2] Add unit tests for many TL2 components 
These tests mostly use the Fuzzer and RAMModel to check that adapters
correctly handle randomly generated legal traffic.
 2016-09-28 18:02:04 -07:00
Henry Cook
81123f84c9 [tilelink2] Make map generation in RRTests a def so that multiple RRTests can be instantiated as part of the same unit test suite. (#356) 2016-09-27 18:06:21 -07:00
Howard Mao
c45cc76cef Get rid of remaining MemIO code 
The only thing we were still using it for was for the MIFDataBits
and MIFTagBits parameters. We replace these with EdgeDataBits and
EdgeIDBits.
 2016-09-27 16:28:17 -07:00