On 4/28/21 12:53 AM, Richard Biener wrote:
On Wed, Apr 28, 2021 at 1:30 AM Martin Sebor via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
The free_lang_data pass is defined entirely in tree.c. Its code
changes only rarely (only 13% commits to tree.c), and unlike
the rest of tree.c, is even more rarely read. The pass is also
right in the middle of tree.c, surrounded by various utility
functions many of which do tend to be frequently referenced (IME),
making the rest of the code harder to find and navigate in than
it should be. The pass contributes nearly 1300 lines of code to
the already quite sizable tree.c (over 16,000 LOC).
To help alleviate some of these problems the attached change moves
the free_lang_data code to its own file where it's easier to find
and work with separately from the rest of tree.c. There are no
functional changes.
Tested on x86_64-linux.
Since it's an IPA pass can you please name it ipa-free-lang-data.cc
(.cc!)?
Okay.
There are now a few functions exported from this, please
consider retaining those in tree.c (I spotted assign_assembler_name_if_needed,
but there may be others),
I didn't notice this function. Thanks for pointing out! I've
moved it back to tree.c, along with the static inline
need_assembler_name_p. Those were the only two.
otherwise policy dictates we need a
ipa-free-lang-data.h and adjust #includes at users.
OK with that changes.
Committed after retesting in g:2de7c792569.
Martin
Thanks,
Richard.
Martin
