Playing with a Fedora chroot & Ion on the HiFive Unleashed | RISCY BUSINESS Episode Guide

0:13Begin to recap and set the stage for the day

🗩

0:13Begin to recap and set the stage for the day

🗩

0:13Begin to recap and set the stage for the day

🗩

0:17riskyfive \o/

🗪

0:17riskyfive \o/

🗪

0:17riskyfive \o/

🗪

0:39riskyfive Hmm, I can't see my own text

🗪

0:39riskyfive Hmm, I can't see my own text

🗪

0:39riskyfive Hmm, I can't see my own text

🗪

1:19riscyfive Ah, any feedback on the podcast? Boring? Confusing?

🗪

1:19riscyfive Ah, any feedback on the podcast? Boring? Confusing?

🗪

1:19riscyfive Ah, any feedback on the podcast? Boring? Confusing?

🗪

4:28Plug riscyfive's podcast 'Why So Anemic'¹

4:42riskyfive Well, I want(ed) to keep my mind open. If I had found out (or I later find out) that the OOP-style modeling was the best option then so be it. I just want to get at the truth, I don't have a camp

🗪

4:42riskyfive Well, I want(ed) to keep my mind open. If I had found out (or I later find out) that the OOP-style modeling was the best option then so be it. I just want to get at the truth, I don't have a camp

🗪

4:42riskyfive Well, I want(ed) to keep my mind open. If I had found out (or I later find out) that the OOP-style modeling was the best option then so be it. I just want to get at the truth, I don't have a camp

🗪

5:30riskyfive People at DConf 2016 thought I was Italian, due to the way I talk

🗪

5:30riskyfive People at DConf 2016 thought I was Italian, due to the way I talk

🗪

5:30riskyfive People at DConf 2016 thought I was Italian, due to the way I talk

🗪

6:11Determine to continue doing stuff with the HiFive Unleashed, noting the possibility from the SiFive Forums to install Fedora²

10:49Explain the process of installing Fedora onto the HiFive Unleashed,³ noting the block size in Arch Linux's example dd invocation⁴

15:42riskyfive By the way, you probably don't have video on your RISC-V system. SiFive used an adapter FPGA board for their demo. I wanted to share with you and your viewers that you can buy the Nexys Video (the bigger one recommended for Bitwise) using an academic discount. I ordered one. We can then later collaborate to get HDMI output using one, or something like that. I can help you with that

🗪

15:42riskyfive By the way, you probably don't have video on your RISC-V system. SiFive used an adapter FPGA board for their demo. I wanted to share with you and your viewers that you can buy the Nexys Video (the bigger one recommended for Bitwise) using an academic discount. I ordered one. We can then later collaborate to get HDMI output using one, or something like that. I can help you with that

🗪

15:42riskyfive By the way, you probably don't have video on your RISC-V system. SiFive used an adapter FPGA board for their demo. I wanted to share with you and your viewers that you can buy the Nexys Video (the bigger one recommended for Bitwise) using an academic discount. I ordered one. We can then later collaborate to get HDMI output using one, or something like that. I can help you with that

🗪

17:02Continue to explain the dd part of the installation process

17:44riskyfive You can even add MP4 encoding in the FPGA itself. No need for OBS

🗪

17:44riskyfive You can even add MP4 encoding in the FPGA itself. No need for OBS

🗪

17:44riskyfive You can even add MP4 encoding in the FPGA itself. No need for OBS

🗪

18:38Continue to explain the installation process

19:08riskyfive You need more Patreons. I'll try to promote you⁵^,6

🗪

19:08riskyfive You need more Patreons. I'll try to promote you⁵^,6

🗪

19:08riskyfive You need more Patreons. I'll try to promote you⁵^,6

🗪

20:49Finish up explaining the setup process, including building the Ion compiler from bitwise and having it run out of the box!

23:00Don the antistatic wristband and plug in the HiFive Unleashed

🗹

23:00Don the antistatic wristband and plug in the HiFive Unleashed

🗹

23:00Don the antistatic wristband and plug in the HiFive Unleashed

🗹

24:03riskyfive Where are you plugging the band?

🗪

24:03riskyfive Where are you plugging the band?

🗪

24:03riskyfive Where are you plugging the band?

🗪

25:03riskyfive The "duct tape" method. Good engineering

🗪

25:03riskyfive The "duct tape" method. Good engineering

🗪

25:03riskyfive The "duct tape" method. Good engineering

🗪

25:23Read 5.3. Knobs on the HiFive Unleashed⁷

27:02riskyfive I've worked with $5K boards at Siemens without grounding and they never exhibited problems. After that I never cared too much

🗪

27:02riskyfive I've worked with $5K boards at Siemens without grounding and they never exhibited problems. After that I never cared too much

🗪

27:02riskyfive I've worked with $5K boards at Siemens without grounding and they never exhibited problems. After that I never cared too much

🗪

28:47riskyfive That actually was very surprising because I kept getting shocked due to the static discharges at that laboratory (when touching the server racks). But the hardware itself never complained

🗪

28:47riskyfive That actually was very surprising because I kept getting shocked due to the static discharges at that laboratory (when touching the server racks). But the hardware itself never complained

🗪

28:47riskyfive That actually was very surprising because I kept getting shocked due to the static discharges at that laboratory (when touching the server racks). But the hardware itself never complained

🗪

30:25riskyfive No, there was a ground. But we were busy and lazy! Maybe that was the trick: I kept discharging myself unintentionally!

🗪

30:25riskyfive No, there was a ground. But we were busy and lazy! Maybe that was the trick: I kept discharging myself unintentionally!

🗪

30:25riskyfive No, there was a ground. But we were busy and lazy! Maybe that was the trick: I kept discharging myself unintentionally!

🗪

30:42Plug in and power up the board

🗹

30:42Plug in and power up the board

🗹

30:42Plug in and power up the board

🗹

31:53riskyfive Yeah, I pretty much only have grounded plugs in the kitchen

🗪

31:53riskyfive Yeah, I pretty much only have grounded plugs in the kitchen

🗪

31:53riskyfive Yeah, I pretty much only have grounded plugs in the kitchen

🗪

32:18Connect to the HiFive Unleashed and setup the environment

🗹

32:18Connect to the HiFive Unleashed and setup the environment

🗹

32:18Connect to the HiFive Unleashed and setup the environment

🗹

36:45riskyfive It seems the cool kids use tmux these days

🗪

36:45riskyfive It seems the cool kids use tmux these days

🗪

36:45riskyfive It seems the cool kids use tmux these days

🗪

37:09Delete our test files related to ion

🗹

37:09Delete our test files related to ion

🗹

37:09Delete our test files related to ion

🗹

40:01riskyfive Try: $ shopt | grep checkwinsize

🗪

40:01riskyfive Try: $ shopt | grep checkwinsize

🗪

40:01riskyfive Try: $ shopt | grep checkwinsize

🗪

41:05Compile ion, run generate_test.py, show the resulting Ion code in test3.ion, run ion on that file to produce test3.c, and compile that test3.c with gcc

🏃

41:05Compile ion, run generate_test.py, show the resulting Ion code in test3.ion, run ion on that file to produce test3.c, and compile that test3.c with gcc

🏃

41:05Compile ion, run generate_test.py, show the resulting Ion code in test3.ion, run ion on that file to produce test3.c, and compile that test3.c with gcc

🏃

43:54Point out that qemu⁸ may provide the closest means to follow along, without having to invest in a HiFive Unleashed

🗩

43:54Point out that qemu⁸ may provide the closest means to follow along, without having to invest in a HiFive Unleashed

🗩

43:54Point out that qemu⁸ may provide the closest means to follow along, without having to invest in a HiFive Unleashed

🗩

45:30Try to run ion on test2.ion and then gcc on the resulting test2.c file

🏃

45:30Try to run ion on test2.ion and then gcc on the resulting test2.c file

🏃

45:30Try to run ion on test2.ion and then gcc on the resulting test2.c file

🏃

46:54Try to run ion on test1.ion and then gcc on the resulting test1.c file, to see printouts and input totally working

🏃

46:54Try to run ion on test1.ion and then gcc on the resulting test1.c file, to see printouts and input totally working

🏃

46:54Try to run ion on test1.ion and then gcc on the resulting test1.c file, to see printouts and input totally working

🏃

48:39Read through test1.ion

51:45Run test1 to provide the first demo of Ion code compiled to and running on RISC-V

🏃

51:45Run test1 to provide the first demo of Ion code compiled to and running on RISC-V

🏃

51:45Run test1 to provide the first demo of Ion code compiled to and running on RISC-V

🏃

52:50Note Miblo's thought to use readelf while he wrote annotations for Bitwise

🗩

52:50Note Miblo's thought to use readelf while he wrote annotations for Bitwise

🗩

52:50Note Miblo's thought to use readelf while he wrote annotations for Bitwise

🗩

55:12Run objdump on test2

🏃

55:12Run objdump on test2

🏃

55:12Run objdump on test2

🏃

55:44Determine to run objdump on hello

🗩

55:44Determine to run objdump on hello

🗩

55:44Determine to run objdump on hello

🗩

56:51riskyfive I use readelf all the time for my empire RISC-V port

🗪

56:51riskyfive I use readelf all the time for my empire RISC-V port

🗪

56:51riskyfive I use readelf all the time for my empire RISC-V port

🗪

57:03Continue to figure out how to unfreeze the screen

🗹

57:03Continue to figure out how to unfreeze the screen

🗹

57:03Continue to figure out how to unfreeze the screen

🗹

57:57riskyfive Ctrl-q?

🗪

57:57riskyfive Ctrl-q?

🗪

57:57riskyfive Ctrl-q?

🗪

58:21Note the annoyance of the XOFF ASCII character and screen's default keybinding for it

💢

58:21Note the annoyance of the XOFF ASCII character and screen's default keybinding for it

💢

58:21Note the annoyance of the XOFF ASCII character and screen's default keybinding for it

💢

59:39riskyfive Yeah, very annoying. Arguably it's just it not ignoring the ASCII characters (XON / XOFF) that you actually typed

🗪

59:39riskyfive Yeah, very annoying. Arguably it's just it not ignoring the ASCII characters (XON / XOFF) that you actually typed

🗪

59:39riskyfive Yeah, very annoying. Arguably it's just it not ignoring the ASCII characters (XON / XOFF) that you actually typed

🗪

1:00:30Copy hello into the (Fedora) chroot environment

🗹

1:00:30Copy hello into the (Fedora) chroot environment

🗹

1:00:30Copy hello into the (Fedora) chroot environment

🗹

1:01:06riskyfive Yeah, but when you type Ctrl-something it emits certain ASCII characters that are control characters, like stop transmitting (freeze)

🗪

1:01:06riskyfive Yeah, but when you type Ctrl-something it emits certain ASCII characters that are control characters, like stop transmitting (freeze)

🗪

1:01:06riskyfive Yeah, but when you type Ctrl-something it emits certain ASCII characters that are control characters, like stop transmitting (freeze)

🗪

1:03:35Run objdump on our elf64-littleriscv hello

🏃

1:03:35Run objdump on our elf64-littleriscv hello

🏃

1:03:35Run objdump on our elf64-littleriscv hello

🏃

1:04:50Consider trying the hula RISC-V "Hello, world!" on our HiFive Unleashed⁹

1:05:37riskyfive That's just what the ASCII character 19 means: stop transmitting. And when you type Ctrl-s you type that character. The real question is, who thought it was a good idea to transmit control information in-band with the content. That is, if you cat a .txt file that contains that character you're screwed

🗪

1:05:37riskyfive That's just what the ASCII character 19 means: stop transmitting. And when you type Ctrl-s you type that character. The real question is, who thought it was a good idea to transmit control information in-band with the content. That is, if you cat a .txt file that contains that character you're screwed

🗪

1:05:37riskyfive That's just what the ASCII character 19 means: stop transmitting. And when you type Ctrl-s you type that character. The real question is, who thought it was a good idea to transmit control information in-band with the content. That is, if you cat a .txt file that contains that character you're screwed

🗪

1:07:14riskyfive You see, it used to be that terminals had dedicated wires for control. But then $ happened

🗪

1:07:14riskyfive You see, it used to be that terminals had dedicated wires for control. But then $ happened

🗪

1:07:14riskyfive You see, it used to be that terminals had dedicated wires for control. But then $ happened

🗪

1:07:31Rant on screen having kept legacy baggage around

💢

1:07:31Rant on screen having kept legacy baggage around

💢

1:07:31Rant on screen having kept legacy baggage around

💢

1:09:31Continue to compare the objdump of hello with hula¹⁰

1:10:02riskyfive With dedicated wires you don't need in-band control characters

🗪

1:10:02riskyfive With dedicated wires you don't need in-band control characters

🗪

1:10:02riskyfive With dedicated wires you don't need in-band control characters

🗪

1:10:43riskyfive No, there was a message missing in between

🗪

1:10:43riskyfive No, there was a message missing in between

🗪

1:10:43riskyfive No, there was a message missing in between

🗪

1:10:52Compare hula's "write" code with the hello objdump

1:11:55riskyfive The Ctrl-z was an example of how the shell has the same problem

🗪

1:11:55riskyfive The Ctrl-z was an example of how the shell has the same problem

🗪

1:11:55riskyfive The Ctrl-z was an example of how the shell has the same problem

🗪

1:13:11Note that the hello objdump may not show scall

1:14:08riskyfive I agree it's braindead. But just think this: the initial Ctrl-a just means that you want the sequence Ctrl-a q to pass as a Ctrl-q to the target process. It's not a list of Ctrl-a commands, it just a generic escape

🗪

1:14:08riskyfive I agree it's braindead. But just think this: the initial Ctrl-a just means that you want the sequence Ctrl-a q to pass as a Ctrl-q to the target process. It's not a list of Ctrl-a commands, it just a generic escape

🗪

1:14:08riskyfive I agree it's braindead. But just think this: the initial Ctrl-a just means that you want the sequence Ctrl-a q to pass as a Ctrl-q to the target process. It's not a list of Ctrl-a commands, it just a generic escape

🗪

1:14:34riskyfive scall -> ecall

🗪

1:14:34riskyfive scall -> ecall

🗪

1:14:34riskyfive scall -> ecall

🗪

1:15:14riskyfive It's the old name

🗪

1:15:14riskyfive It's the old name

🗪

1:15:14riskyfive It's the old name

🗪

1:15:16Wonder if the objdump is not using system calls, and that it's a lot bigger than hula

1:16:19riskyfive Notice: disassembly of XXX (not .text) section

🗪

1:16:19riskyfive Notice: disassembly of XXX (not .text) section

🗪

1:16:19riskyfive Notice: disassembly of XXX (not .text) section

🗪

1:16:42Hunt for a7 in the objdump because that is where you put the syscall number in RISC-V

1:18:11riskyfive You can do a readelf of the binary in your main system

🗪

1:18:11riskyfive You can do a readelf of the binary in your main system

🗪

1:18:11riskyfive You can do a readelf of the binary in your main system

🗪

1:18:27Consult man readelf

1:19:25Run readelf -a hello

🏃

1:19:25Run readelf -a hello

🏃

1:19:25Run readelf -a hello

🏃

1:20:21Read through the readelf of hello

1:21:40riskyfive Maybe just try GDB with the binary?

🗪

1:21:40riskyfive Maybe just try GDB with the binary?

🗪

1:21:40riskyfive Maybe just try GDB with the binary?

🗪

1:22:13Continue to read through the readelf of hello

1:22:31riskyfive That's why you need the JTAG port

🗪

1:22:31riskyfive That's why you need the JTAG port

🗪

1:22:31riskyfive That's why you need the JTAG port

🗪

1:22:46Return to the objdump of hello

1:23:42riskyfive The system calls are probably wrapped in libc functions

🗪

1:23:42riskyfive The system calls are probably wrapped in libc functions

🗪

1:23:42riskyfive The system calls are probably wrapped in libc functions

🗪

1:23:55Notice the mentions of "libc" in the hello objdump

1:24:49riskyfive I think the ligc_start_main is just a red herring. Notice the offset is different between different __libc_start_main lines

🗪

1:24:49riskyfive I think the ligc_start_main is just a red herring. Notice the offset is different between different __libc_start_main lines

🗪

1:24:49riskyfive I think the ligc_start_main is just a red herring. Notice the offset is different between different __libc_start_main lines

🗪

1:25:58Notice "libc_start_main" in the hello objdump

1:27:59Consult the objdump of test1, to see that its symbols are correctly resolved

1:30:39Call it a day

🗩

1:30:39Call it a day

🗩

1:30:39Call it a day

🗩

RISCY BUSINESS

Keyboard Navigation

Global Keys

Menu toggling

In-Menu Movement

Quotes and References Menus

Quotes, References and Credits Menus

Filter Menu

Filter and Link Menus

Credits Menu