Enable programmable logic support

Goal

In this tutorial you will

  • Expand Vivado project from Create minimalist Vivado project for Leopard DPU with support for Programmable Logic

  • Build bitstream with two UART peripherals connected together

A bit of background

Leopard DPU configuration that you created in Create minimalist Vivado project tutorial has support only for Processing System. Usage of Programmable Logic part of Zynq UltraScale+ requires following capabilities from Processing System side:

  • Clock for programmable logic

  • Interrupt lines between Processing System and Programmable Logic

  • Memory interface between Processing System and Programmable Logic

Zynq UltraScale+ IP block (the same block that you used to configure base settings like DDR memory or I/O pins) configures these options. All Vivado projects describing different “content” of Programmable Logic must use the same base settings. Vivado provides preset mechanism to transfer processing system configuration between projects.

../../../_images/vivado_presets_share.drawio1.png

Fig. 8 Multiple Vivado projects sharing Processing System configuration using presets and producing different .xsa files.

Building content for programmable logic involves putting different IP blocks on block design, connecting them together using memory interfaces, interrupts, and clocks. To be usable from Linux distribution, IPs connect to “extension points” provided by Processing System.

Vivado stores content for Programmable Logic in .xsa file in form called bitstream. Using running Linux operating system you can use FPGA Manager to load different bitstreams to Programmable Logic.

Prerequisites

Provided outputs

Following files (Tutorial files) are associated with this tutorial:

  • Leopard/Zero-to-hero/03 Enable programmable logic support/leopard-minimalistic-with-pl.tcl - Vivado preset for Leopard with enabled PL support

  • Leopard/Zero-to-hero/03 Enable programmable logic support/leopard-minimalistic-pl-base.xsa - Base XSA file with enabled PL support

  • Leopard/Zero-to-hero/03 Enable programmable logic support/leopard-double-uart.xsa - Double UART XSA file

  • Leopard/Zero-to-hero/03 Enable programmable logic support/boot-common.bin - Boot firmware for Leopard

  • Leopard/Zero-to-hero/03 Enable programmable logic support/nominal-image-leopard-dpu.rootfs.cpio.gz.u-boot - Root filesystem for Leopard

  • Leopard/Zero-to-hero/03 Enable programmable logic support/Image - Linux kernel

  • Leopard/Zero-to-hero/03 Enable programmable logic support/system.dtb - Device tree

Use these files if you want to skip building bitstream or Yocto distribution by yourself.

Enable programmable logic support Vivado

  1. Open Leopard DPU project from Create minimalist Vivado project in Vivado

  2. Use Open Block Design option to open top_bd block design

  3. Customize Zynq UltraScale+ block by double-clicking on it

    • Enable PL to PS interrupts IRQ0[0-7]

    • Enable PS-PL Master interface AXI HPM0 FPD

    • Enable PL-PS Slave interfaces

      • Enable AXI HPC0 FPD

      • Enable AXI HPC1 FPD

      • Enable AXI LPD, set Data Width to 32 bits

    • Enable Fabric Reset Enable

    • Set Number of Fabric Resets to 1

    • Enable PL fabric clock in Output clocks tab

      • Enable PL0 and set it to 100MHz

  4. In top_bd block design connect maxihpm0_fpd_aclk to pl0_clk

  5. In top_bd block design connect saxihpc0_fpd_aclk to pl0_clk

  6. In top_bd block design connect saxihpc1_fpd_aclk to pl0_clk

  7. In top_bd block design connect saxi_lpd_aclk to pl0_clk

  8. At this point block design should contain single IP block with each clock connected to pl0_clk

    ../../../_images/leopard_pl_support_enabled.png

    Fig. 9 Block design with Zynq UltraScale+ IP block configured to support Programmable Logic

  9. Open customization of Zynq UltraScale+ IP block and export preset by selecting Presets ‣ Save configuration

    • Use leopard-minimalistic-with-pl as preset name

    • Save to leopard-minimalistic-with-pl.tcl file

  10. Generate bitstream

  11. Export hardware without bitstream. Use leopard-minimalistic-pl-base.xsa for output file name.

Note

Selected Zynq UltraScale+ configuration covers needs of programmable logic content in this tutorial and next ones.

Create double UART bitstream Vivado

  1. Start Vivado and create new project. In new project wizard select following options:

    • Project type: RTL Project

      • Select Don’t specify sources at this time

      • Don’t select Project is an extensible Vitis platform

    • Part: xczu9eg-ffvc900-1L-i

  2. Create top-level block design by using Create Block Design in Flow Navigator. Use double_uart_bd as name.

  3. In block design diagram editor add Zynq UltraScale+ MPSoC IP block.

  4. Start customization of Zynq UltraScale+ MPSoC IP block by double-clicking on it.

    1. Apply previously exported preset by selecting Presets ‣ Apply configuration and select leopard-minimalistic-with-pl.tcl file.

  5. In double_uart_bd block design connect maxihpm0_fpd_aclk to pl0_clk.

  6. In double_uart_bd block design connect saxihpc0_fpd_aclk to pl0_clk

  7. In double_uart_bd block design connect saxihpc1_fpd_aclk to pl0_clk

  8. In double_uart_bd block design connect saxi_lpd_aclk to pl0_clk

  9. Place two AXI Uartlite IPs on block design

  10. Cross-connect UARTs by connecting axu_uartlite1 TX to axu_uartlite0 RX and vice versa.

  11. Click Run connection automation and let Vivado instantiate necessary interconnects and resets.

  12. Add Concat IP block

  13. Connect dout pin of Concat block to pl_ps_irq pin of Zynq UltraScale+ block

  14. Connect interrupt pin of axi_uartlite0 to In0 of Concat block

  15. Connect interrupt pin of axi_uartlite1 to In1 of Concat block

  16. Final block design should look like this:

    ../../../_images/leopard_double_uart_bd.png

    Fig. 10 Block design with double UARTs connected together and available to Processing System

  17. In Sources view select Design Sources ‣ double_uart_bd and click Create HDL Wrapper` in context menu. Use Let Vivado manage wrapper and auto-update option.

  18. Generate bitstream

  19. Export hardware including bitstream to file leopard-double-uart.xsa

Enable programmable logic support in boot firmware Yocto

  1. Add leopard-minimalistic-pl-base.xsa to sources/meta-local/recipes-bsp/hdf/external-hdf/ directory.

  2. Modify sources/meta-local/recipes-bsp/hdf/external-hdf_%.bbappend to use new XSA file.

    FILESEXTRAPATHS:prepend := "${THISDIR}/${PN}:"

HDF_BASE = "file://"
HDF_PATH = "leopard-minimalistic-pl-base.xsa"

Add double UART bitstream to Linux distribution Yocto

Note

If necessary, re-enable Yocto environment using

machine:~/leopard-linux-1$ source sources/poky/oe-init-build-env ./build

  1. Create directory sources/meta-local/recipes-example/bitstreams/double-uart/ and copy leopard-double-uart.xsa to it.

  2. Create new recipe sources/meta-local/recipes-example/bitstreams/double-uart.bb that will install bitstream with double UART.

    LICENSE = "CLOSED"

inherit bitstream

SRC_URI += "file://leopard-double-uart.xsa"
BITSTREAM_HDF_FILE = "${WORKDIR}/leopard-double-uart.xsa"

  3. Create append for leopard-all recipe

    machine:~/leopard-linux-1/build$ recipetool newappend ../sources/meta-local/ nominal-image
NOTE: Starting bitbake server...
WARNING: The ZynqMP pmu-rom is not enabled, qemu may not be able to emulate a ZynqMP system without it. To enable this you must add 'xilinx' to the LICENSE_FLAGS_ACCEPTED to indicate you accept the software license.
Loading cache: 100% |#############################################################################################################################################################################| Time: 0:00:00
Loaded 3317 entries from dependency cache.
Parsing recipes: 100% |###########################################################################################################################################################################| Time: 0:00:00
Parsing of 2006 .bb files complete (2004 cached, 2 parsed). 3318 targets, 570 skipped, 0 masked, 0 errors.
WARNING: No bb files in default matched BBFILE_PATTERN_meta-kp-classes '^~/leopard-linux-1/sources/meta-kp-classes/meta-kp-classes/'

Summary: There was 1 WARNING message.
~/leopard-linux-1/sources/meta-local/recipes-leopard/images/nominal-image.bbappend

  4. Add new packages into Linux image by editing sources/meta-local/recipes-leopard/images/nominal-image.bbappend

    IMAGE_INSTALL += "\
    fpga-manager-script \
    double-uart \
"

  5. Build firmware and image

    machine:~/leopard-linux-1/build$ bitbake leopard-all

  6. Prepare build artifacts for transfer to EGSE Host

    machine:~/leopard-linux-1/build$ mkdir -p ../egse-host-transfer
machine:~/leopard-linux-1/build$ cp tmp/deploy/images/leopard-dpu/bootbins/boot-common.bin ../egse-host-transfer
machine:~/leopard-linux-1/build$ cp tmp/deploy/images/leopard-dpu/system.dtb ../egse-host-transfer
machine:~/leopard-linux-1/build$ cp tmp/deploy/images/leopard-dpu/nominal-image-leopard-dpu.rootfs.cpio.gz.u-boot ../egse-host-transfer
machine:~/leopard-linux-1/build$ cp tmp/deploy/images/leopard-dpu/Image ../egse-host-transfer

  7. Transfer content of egse-host-transfer directory to EGSE Host and place it in /var/tftp/tutorial directory

Loading double UART bitstream on DPU EGSE Host

  1. Verify that all necessary artifacts are present on EGSE Host:

    customer@egse-host:~$ ls -lh /var/tftp/tutorial
total 54M
-rw-rw-r-- 1 customer customer  21M Jul 16 07:35 Image
-rw-rw-r-- 1 customer customer 1.6M Jul 16 07:35 boot-common.bin
-rw-rw-r-- 1 customer customer  41M Jul 16 07:35 nominal-image-leopard-dpu.rootfs.cpio.gz.u-boot
-rw-rw-r-- 1 customer customer  39K Jul 16 07:35 system.dtb

    Note

    Exact file size might differ a bit but they should be in the same range (for example nominal-image-leopard-dpu.rootfs.cpio.gz.u-boot shall be about ~40MB)

  2. Ensure that Leopard is powered off

    customer@egse-host:~$ sml power off
Powering off...Success

  3. Open second SSH connection to EGSE Host and start minicom to observe boot process

    customer@egse-host:~$ minicom -D /dev/sml/leopard-pn1-uart

    Leave this terminal open and get back to SSH connection used in previous steps.

  4. Power on Leopard

    customer@egse-host:~$ sml power on
Powering on...Success

  5. Write boot firmware to DPU boot flash

    customer@egse-host:~$ sml boot-flash write --nor-memory nor1 0 /var/tftp/tutorial/boot-common.bin
Uploading   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00 48.6 MB/s
Erasing     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00 553.3 kB/s
Programming ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00 13.5 kB/s

  6. Power on Processing Node 1

    customer@egse-host:~$ sml pn1 power on --nor-memory nor1
Powering on processing node Node1...Success

  7. DPU boot process should be visible in minicom terminal

  8. Log in to DPU using root user

    leopard login: root
root@leopard:~#

  9. Load double UART bitstream

    root@leopard:~# fpgautil -o /lib/firmware/double-uart/overlay.dtbo
[   17.334051] fpga_manager fpga0: writing double-uart/bitstream.bit.bin to Xilinx ZynqMP FPGA Manager
[   17.478795] OF: overlay: WARNING: memory leak will occur if overlay removed, property: /fpga-full/firmware-name
[   17.488941] OF: overlay: WARNING: memory leak will occur if overlay removed, property: /fpga-full/resets
[   17.498582] OF: overlay: WARNING: memory leak will occur if overlay removed, property: /__symbols__/afi0
[   17.508081] OF: overlay: WARNING: memory leak will occur if overlay removed, property: /__symbols__/axi_uartlite_0
[   17.518445] OF: overlay: WARNING: memory leak will occur if overlay removed, property: /__symbols__/axi_uartlite_1
[   17.532846] a0000000.serial: ttyUL0 at MMIO 0xa0000000 (irq = 45, base_baud = 0) is a uartlite
[   17.543564] uartlite a0000000.serial: Runtime PM usage count underflow!
[   17.553041] a0010000.serial: ttyUL1 at MMIO 0xa0010000 (irq = 46, base_baud = 0) is a uartlite
[   17.563853] uartlite a0010000.serial: Runtime PM usage count underflow!
root@leopard:~#

    Note

    Despite warnings UARTs in bitstream will still function correctly

  10. Verify presence of two new UART devices

    root@leopard:~# ls -l /dev/ttyUL*
crw-rw----    1 root     dialout   204, 187 Sep 20 11:23 /dev/ttyUL0
crw-rw----    1 root     dialout   204, 188 Sep 20 11:23 /dev/ttyUL1

  11. Start receiving data from /dev/ttyUL0 in background

    root@leopard:~# cat /dev/ttyUL0 &

    cat process will be running in background allowing you to enter another command in the same terminal. Output from cat (data received from UART) and your commands will mix in terminal.

  12. Write something to second UART:

    root@leopard:~# echo "Hello from UART1" > /dev/ttyUL1
Hello from UART1
root@leopard:~#

    Text Hello from UART1 is coming from cat running in background.

Summary

In this tutorial, you enabled usage of Programmable Logic part of Zynq UltraScale+ device. As an example, you added bitstream with two UARTs connected together. After rebuilding Yocto project, you used FPGA Manager to load bitstream dynamically and used newly added devices.