Application Acceleration With FPGAs
Programmable Acceleration Cards (PACs), DCP, FPGA AI Suite, Software Stack, and Reference Designs
475 Discussions

Problem when running DMA_AFU on ASE simulation.

spang2
Beginner
1,005 Views

I am using 1.1 production version and using DMA_AFU from sample to build ASE simulation.

 

From the simluation script, I can invoke the simulation, the log file is:

  [SIM] Simulator started...

#  [SIM] +CONFIG /home/xxxxx/intel/dma_afu/build_sim/ase.cfg file found !

#  [SIM] +SCRIPT /home/xxxxx/intel/dma_afu/build_sim/ase_regress.sh file found !

#  [SIM] PID of simulator is 82746

#  [SIM] Reading /home/xxxxx/intel/dma_afu/build_sim/ase.cfg configuration file

#  [SIM] ASE was started in Mode 3 (Server-Client with Sw SIMKILL (long runs)

#  [SIM] ASE Mode: Server-Client mode with SW SIMKILL (long runs)

#  [SIM] Inactivity kill-switch   ... DISABLED

#  [SIM] Reuse simulation seed   ... ENABLED

#  [SIM] ASE Seed          ... 1234

#  [SIM] ASE Transaction view    ... DISABLED

#  [SIM] User Clock Frequency    ... 312.500000 MHz, T_uclk = 3200 ps

#  [SIM] Amount of physical memory ... 128 GB

#  [SIM] Current Directory located at =>

#  [SIM] /home/xxxxx/intel/dma_afu/build_sim/work

#  [SIM] Creating Messaging IPCs...

#  [SIM] Calculating memory map...

#  [SIM]     System memory range => 0x0000000000000000-0x0000001fffffffff | 0~128 GB

#  [SIM] Information about allocated buffers => workspace_info.log

#  [SIM] ASE running with seed =>    1234

#  [SIM] Sending initial reset...

#  [SIM] Transaction Logger started

#  [SIM] Protocol Checker initialized

#  [SIM] ASE lock file .ase_ready.pid written in work directory

#  [SIM] ** ATTENTION : BEFORE running the software application **

#  [SIM] Set env(ASE_WORKDIR) in terminal where application will run (copy-and-paste) =>

#  [SIM] $SHELL  | Run:

#  [SIM] ---------+---------------------------------------------------

#  [SIM] bash/zsh | export ASE_WORKDIR=/home/xxxxx/intel/dma_afu/build_sim/work

#  [SIM] tcsh/csh | setenv ASE_WORKDIR /home/xxxxx/intel/dma_afu/build_sim/work

#  [SIM] For any other $SHELL, consult your Linux administrator

#  [SIM]

#  [SIM] Ready for simulation...

#  [SIM] Press CTRL-C to close simulator...

#  [SIM] Session requested by PID = 82796

------------------------------------------------------------------------------

However, from the sample sw side, when I executed the application

I got the following error:

 [APP] ASE Session Directory located at =>

 [APP] /home/xxxxx/intel/dma_afu/build_sim/work

 [APP] ASE found a stale Application lock with PID = 82096 -- this will be removed

 [APP] Deleted the existing app_lock.pid with Stale pid

 [APP] Initializing simulation session ...

 [APP]

 [APP] Issuing Soft Reset...

 [APP] Session started

 [APP] ASE Capabilities: Base Intr

 [APP] Waiting till session ID is created by ASE ...  [APP] DONE

 [APP] Creating MMIO ...

 [APP] Attempting to open a shared memory...

 [APP] SUCCESS

 [APP] AFU MMIO Virtual Base Address = 0x7f5a99a74000

 [APP] Creating UMAS ...

 [APP] Attempting to open a shared memory...

 [APP] SUCCESS

 [APP] UMAS Virtual Base address = 0x7f5a988a4000

 [APP] Starting MMIO Read Response watcher ...

 [APP] SUCCESS

 [APP] Starting UMsg watcher ...

 [APP] SUCCESS

 [APP] Application received a  [APP] SIGBUS

 [APP]

[bt] Execution Backtrace:

 [APP] [bt] #1 /lib64/libc.so.6(+0x36280) [0x7f5a98cf6280]

??:0

 [APP] [bt] #2 /lib64/libc.so.6(+0x15ae88) [0x7f5a98e1ae88]

??:0

 [APP] [bt] #3 /lib64/libopae-c-ase.so.1(ase_memcpy_s+0x5f) [0x7f5a9909356f]

??:0

 [APP] [bt] #4 /lib64/libopae-c-ase.so.1(umsg_watcher+0x7f) [0x7f5a9909393f]

??:0

 [APP] [bt] #5 /lib64/libpthread.so.0(+0x7dd5) [0x7f5a98aabdd5]

??:0

 [APP] [bt] #6 /lib64/libc.so.6(clone+0x6d) [0x7f5a98dbdead]

----------------------------------------------------------------

The syntax I run the application is :

sudo LD_LIBRARY_PATH=$PWD ASE_WORKDIR=/home/xxxxx/intel/dma_afu/build_sim/work : ./fpga_dma_test 1

 

Can anyone tell me how to fix the problem?

 

Thanks a lot.

Sunny

 

 

 

 

0 Kudos
3 Replies
Divya_V_Intel
Employee
712 Views

​Hi Sunny,

The 1.1 dma_afu has a couple of known bugs. Would you be open to upgrading to the 1.2 production release to see if the issue persists, please? Here are instructions on getting started with the 1.2 release - https://www.intel.com/content/www/us/en/programmable/products/boards_and_kits/dev-kits/altera/acceleration-card-arria-10-gx/getting-started.html.

If the issue persists in 1.2, I will replicate it at my end and dig deeper.

Thanks,

Divya, Intel Technical Support

spang2
Beginner
712 Views

Hi Divya,

 

Thanks for your reply. I made several mistake, it is working now.

The mistake I made is:

  1. I run "make" to compile the application. For ASE, I should run "make USE_ASE=1"
  2. To run the application, I used wrong command. I should just run "./fpga_dma_test", instead of "sudo LD_LIBRARY_PATH=$PWD ASE_WORKDIR=/home/xxxxx/intel/dma_afu/build_sim/work : ./fpga_dma_test 1"

 

Thanks,

Sunny

0 Kudos
Divya_V_Intel
Employee
712 Views

​Hi Sunny,

Glad to hear it is working now.

Thanks,

Divya

Reply