Author Archives: Adam Young

About Adam Young

Once upon a time I was an Army Officer, but that was long ago. Now I work as a Software Engineer. I climb rocks, play saxophone, and spend way too much time in front of a computer.

Testing an MCTP Driver via Echo request

Posted on by
Reply

MCTP stands for Management Component Transport Protocol. While it is designed around a server, it is also designed as a network protocol. As such, the Linux implementation makes use of the Socket interface and the Kernel plumbing for dealing with network protocols. To support a new transport mechanism, you implement a new device driver specific to that transport mechanism that implements the struct net_device contract, which includes implementing the functions for struct net_device_ops.

Before I write a full implementation, I want to write something that only echos a packet back to the receiver. This mimics the behavior of the mctp-echo server in the user tools, and can make use of the mctp-req echo client.

Continue reading →

Remote git with ssh

Posted on by
Reply

Working on a different architecture from my Laptop means that I am invariably working on a remote machine. My current development can be done on an Ampere AltraMax machine, which means I have 80 processors available; quite a nice benefit when doing a Linux Kernel compile that can use all of the processors available.

Because the machine is a shared resource out of out lab, I want to make sure I can recreate my work there on another machine; this one could be reclaimed or powered down due to lab priorities. Thus, all my remote work is done in git, and I use the ssh protocol to pull changes from my work server to my laptop fairly regularly…whenever I feel I have valuable work that could be potentially lost.

Continue reading →

Acronym Challenge Programmatic Interface

Posted on by
Reply

How do you know what is inside your computer? There are a couple tools. If the hardware is on the PCI bus, from the command line you can run lspci, which will in turn enumerate the discovered devices on that bus. But what if the hardware is not on the PCI bus? And how does the Kernel discover it in the first place? For the hardware that I have to work with, the answer is that it is enumerated by the Unified Extensible Firmware Interface (UEFI) coded embedded in the device and exposed via the Advanced Configuration and Power Interface (ACPI). This world is full of four letter acronyms. Here are my notes on some of them.

Continue reading →

Keeping build output on one screen

Posted on by
Reply

When a build goes wrong, the amount of error messaging can easily scroll off the screen. Usually the error is on the first line reported. Here’s a couple ways to make it easier to read just the lines you want.

The first is just to grep for the word ‘error.’ This works for gcc:

make 2>&1 | grep error

Which produces

 
/root/devel/mctp_pcc/mctp_pcc.c:178:18: error: ‘name’ undeclared (first use in this function)
/root/devel/mctp_pcc/mctp_pcc.c:178:52: error: ‘fe’ undeclared (first use in this function); did you mean ‘fd’?
/root/devel/mctp_pcc/mctp_pcc.c:179:9: error: ‘ndev’ undeclared (first use in this function); did you mean ‘cdev’?
/root/devel/mctp_pcc/mctp_pcc.c:179:37: error: ‘dev’ undeclared (first use in this function); did you mean ‘cdev’?
/root/devel/mctp_pcc/mctp_pcc.c:182:17: error: ‘rc’ undeclared (first use in this function); did you mean ‘rq’?
/root/devel/mctp_pcc/mctp_pcc.c:187:20: error: ‘idx’ undeclared (first use in this function); did you mean ‘ida’?
/root/devel/mctp_pcc/mctp_pcc.c:190:24: error: ‘STATE_IDLE’ undeclared (first use in this function); did you mean ‘VTIME_IDLE’?
/root/devel/mctp_pcc/mctp_pcc.c:193:34: error: ‘mctp_serial_tx_work’ undeclared (first use in this function)
/root/devel/mctp_pcc/mctp_pcc.c:197:17: error: label ‘free_netdev’ used but not defined
/root/devel/mctp_pcc/mctp_pcc.c:183:17: error: label ‘free_ida’ used but not defined
/root/devel/mctp_pcc/mctp_pcc.c:199:1: error: no return statement in function returning non-void [-Werror=return-type]

A bit of background…all Linux processes have 3 file descriptors (FD) by default. STDIN is for reading in information from other processes. STDIN is file descriptor 0. STDOUT is for, well, STANDARD’ output, or the output that the program is supposed to produce. STDOUT is FD 1. STDERR is for output that is not standard, and is intended for error messages. STDERR is FD 2.. Make and GCC spit their output into STDERR. The trick is to tell the program to redirect standard error into STDOUT via the magic symbol 2>&1. I read this in my head as “two goes into ampersand one.” STDERR is 2, STDOUT is 1.

If you want to get just the first set of line you can use the head command like this:

make 2>&1 | head -10

Which produces

make -C /lib/modules/6.2.0+/build M=/root/devel/mctp_pcc modules
make[1]: Entering directory '/root/devel/linux'
  CC [M]  /root/devel/mctp_pcc/mctp_pcc.o
/root/devel/mctp_pcc/mctp_pcc.c: In function ‘create_mctp_pcc_nnetdev’:
/root/devel/mctp_pcc/mctp_pcc.c:178:18: error: ‘name’ undeclared (first use in this function)
  178 |         snprintf(name, sizeof(name), "mctpipcc%x", fe);
      |                  ^~~~
/root/devel/mctp_pcc/mctp_pcc.c:178:18: note: each undeclared identifier is reported only once for each function it appears in
/root/devel/mctp_pcc/mctp_pcc.c:178:52: error: ‘fe’ undeclared (first use in this function); did you mean ‘fd’?
  178 |         snprintf(name, sizeof(name), "mctpipcc%x", fe);

This gets just the first 10 lines…use a different number if you want a different amount of output

These two options are fairly easy to type and thus don’t really call for scripting. As the amount of filtering gets longer,. you make want to make scripts that build up more complex selection of the output.

If you have control of your makefile, you can use the -Wfatal-errorsflag as is written up here, but I am calling into precanned Makefiles and it drops the flag. To modify Makefiles see this discussion.