Bring iniparser library codebase in line with current version 4.1

[dgsga]

This commit removes deprecated strcpy calls from iniparser and fixes the associated buffer overflow causing afpd to crash when compiled with clang.

[rdmark]

@dgsga Thanks for another round of PRs! I appreciate you taking the time to port over improvements from your fork.

I'm not too familiar with iniparser, but I assume this is autogenerated code here in this PR? Just for my future reference, how was it generated?

[dgsga]

The comments explaining each function are taken verbatim from @ndevilla's iniparser repo on GitHub. Each function was adapted from this repo after comparing the Netatalk iniparser implementation with the original v3 of iniparser.

[dgsga]

@rdmark, I have been thinking about creating a big PR that would bring compatibility with modern Intel / Apple Silicon macOS hosts to the upstream master. What do you think??

[rdmark]

The comments explaining each function are taken verbatim from @ndevilla's iniparser repo on GitHub. Each function was adapted from this repo after comparing the Netatalk iniparser implementation with the original v3 of iniparser.

That makes sense. I realized only now when looking at the commit history that iniparser is a 3rd party library originally. With iniparser being distributed under the MIT license, I wonder if a copyright notice is apt for these source files? Our own COPYING only has GPL2, so perhaps adding this to the header of all iniparser source files is appropriate: https://github.com/ndevilla/iniparser/blob/master/LICENSE

IANAL, but this is my takeaway from a few minutes of researching online what to do when shipping MIT licensed code in a project distributed under the GPL.

BTW, I assume this changeset is proven in your fork? The iniparser readme mentions breaking API changes between v3 and v4.

[rdmark]

I have been thinking about creating a big PR that would bring compatibility with modern Intel / Apple Silicon macOS hosts to the upstream master. What do you think??

I would personally be all for that! Would be handy to run netatalk3 on my M1 MBP (though this is already possible with your fork of course.) Would you mind raising a github ticket for this for tracking?

[dgsga]

The comments explaining each function are taken verbatim from @ndevilla's iniparser repo on GitHub. Each function was adapted from this repo after comparing the Netatalk iniparser implementation with the original v3 of iniparser.

That makes sense. I realized only now when looking at the commit history that iniparser is a 3rd party library originally. With iniparser being distributed under the MIT license, I wonder if a copyright notice is apt for these source files? Our own COPYING only has GPL2, so perhaps adding this to the header of all iniparser source files is appropriate: https://github.com/ndevilla/iniparser/blob/master/LICENSE

IANAL, but this is my takeaway from a few minutes of researching online what to do when shipping MIT licensed code in a project distributed under the GPL.

BTW, I assume this changeset is proven in your fork? The iniparser readme mentions breaking API changes between v3 and v4.

I agree, the license needs to be added as the original code in @ndevilla's repo is released under its terms. Quoting from the license itself:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

So license is added, force-pushed to PR. I have been using the basis of this patch for 2 years on my fork without issue. I recently updated it to add the function explanations and a small amount of missing buffer overflow check code that is present in @ndevilla's repo

[dgsga]

I have been thinking about creating a big PR that would bring compatibility with modern Intel / Apple Silicon macOS hosts to the upstream master. What do you think??

I would personally be all for that! Would be handy to run netatalk3 on my M1 MBP (though this is already possible with your fork of course.) Would you mind raising a github ticket for this for tracking?

Ticket added.

[rdmark]

Perfect, thank you!

[rdmark]

FWIW gcc10 throws a few warnings with the new code that weren't there before:

make[3]: Entering directory '/home/dmark/dev/netatalk/libatalk/iniparser'
  CC       dictionary.lo
  CC       iniparser.lo
iniparser.c: In function ‘atalk_iniparser_line’:
iniparser.c:546:27: warning: passing argument 1 of ‘strstrip’ discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
  546 |     strcpy(line, strstrip(input_line));
      |                           ^~~~~~~~~~
iniparser.c:74:31: note: expected ‘char *’ but argument is of type ‘const char *’
   74 | static char * strstrip(char * s)
      |                        ~~~~~~~^
In function ‘atalk_iniparser_line’,
    inlined from ‘atalk_iniparser_load’ at iniparser.c:678:17:
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
  597 |     free(line);
      |     ^~~~~~~~~~
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
iniparser.c:597:5: warning: attempt to free a non-heap object ‘line’ [-Wfree-nonheap-object]
  CCLD     libiniparser.la
make[3]: Leaving directory '/home/dmark/dev/netatalk/libatalk/iniparser'

[dgsga]

@rdmark, thanks for flagging this. I spotted in your log that line 546 of iniparser.c contains a call to strcpy that should not be there! Checked my files and it appears that I made an error in how I created the PR. Please see the latest PR for the fix. Let me know if you would like to revert the previous commit and then apply a fresh PR or are happy just to apply the fix. Apologies!