linuxkpi: Add hex_dump_to_buffer()
ClosedPublic
Actions

Authored by dumbbell on Jul 26 2025, 3:53 PM.

Details

Reviewers

Group Reviewers

linuxkpi

Commits

rGada63d6b2e5e: linuxkpi: Add hex_dump_to_buffer()
rGb1bef9f7e0e1: linuxkpi: Add hex_dump_to_buffer()

Summary

This function prints a single line of hex dump to the given line buffer.

The implementation relies on lkpi_hex_dump() to format the string. After the formatting, it trims the newline character automatically added by lkpi_hex_dump().

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

dumbbell created this revision.Jul 26 2025, 3:53 PM

Herald added subscribers: emaste, imp. · View Herald TranscriptJul 26 2025, 3:53 PM

dumbbell requested review of this revision.Jul 26 2025, 3:53 PM

It looks like Linux hex_dump_to_buffer does a hex+printable ASCII a la 48 65 6c 6c 6f 2c 20 77 6f 72 6c 64 Hello, world, although I don't think it's incredibly important for our use cases.

In D51559#1178657, @emaste wrote:

It looks like Linux hex_dump_to_buffer does a hex+printable ASCII a la 48 65 6c 6c 6f 2c 20 77 6f 72 6c 64 Hello, world, although I don't think it's incredibly important for our use cases.

Yeah, we do have the ascii flag but it was never implemented (not before my reshuffling either).

sys/compat/linuxkpi/common/src/linux_compat.c
2013	you do not want to return negative I think but 0; otherwise linelen in lkpi_hex_dump may go negative or back to 0; this is likely also a problem with the other callback routines.
2015	this is not correct. written can be > context->linebuflen if the output would have been longer than the buffer size. You need another check here or rather set it accordingly .. hmm also below is wrong to the end of the func. If I am not mistaken: if (written <= (context->linebuflen - 1)) /* terminating \0 */ context->written = written + 1; else context->written = context->linebuflen; context->linebuf += context->written; context->linebuflen -= context->written; return (context->written); But you should double-check that given vsnprintf returns the length without the trailing \0 and I am not sure if you need to count it or not.
2044	This is not needed, is it? lkpi_hex_dump will only do 8/4/2 or 1 in every other case
2060	should we add another bool to lkpi_hex_dump whether to add the (last) \n or not?

This revision now requires changes to proceed.Jul 30 2025, 9:16 AM

dumbbell added a parent revision: D51844: linuxkpi: Add option to skip trailing newline in `lkpi_hex_dump()`.Aug 9 2025, 12:13 PM

dumbbell added a parent revision: D51558: linuxkpi: Avoid trailing whitespaces in lkpi_hex_dump().

dumbbell marked 4 inline comments as done.Aug 9 2025, 12:18 PM

dumbbell added inline comments.

sys/compat/linuxkpi/common/src/linux_compat.c
2013	You are right. I updated D51558 to handle negative return values from the callback.
2015	According to the documentation of `hex_dump_to_buffer()` on Linux, it has the same behaviour as `snprintf()`: it returns the number of bytes it would have written if the destination buffer was large enough, excluding the trailing NUL character. I updated the patch to remove the `linebuflen == 0` check at the beginning and added comments to describe that. Is it clearer or is there still a logic issue?
2044	You are right. I removed that part.
2060	Good idea, I added D51844 to implement that separately.