[00/22] More splitting of dwarf2/read.c

Message ID 20200322184523.28959-1-tom@tromey.com
Headers show
Series
  • More splitting of dwarf2/read.c
Related show

Message

Tom Tromey March 22, 2020, 6:45 p.m.
One goal of mine is to split up the DWARF reader code into many
smaller, more self-contained files.  I think this will make the code
easier to read and to work on; and also let us tease out (and repair)
hidden dependencies.  This latter part is desirable because, in the
longer run, I'd like to parallelize DWARF reading.

So, here is round 2 of splitting.

The main change here is that the macro-reading code is moved to its
own file.  It's almost possible, after this, to scan macros in a
worker thread.  (However, dwz file handling, and the recent move to
sharing a bcache, interfere with this.)

Some smaller utility functions are also moved into other files as
well.

Let me know what you think,
Tom

Comments

Simon Marchi March 24, 2020, 1:34 p.m. | #1
On 2020-03-22 2:45 p.m., Tom Tromey wrote:
> One goal of mine is to split up the DWARF reader code into many

> smaller, more self-contained files.  I think this will make the code

> easier to read and to work on; and also let us tease out (and repair)

> hidden dependencies.  This latter part is desirable because, in the

> longer run, I'd like to parallelize DWARF reading.

> 

> So, here is round 2 of splitting.

> 

> The main change here is that the macro-reading code is moved to its

> own file.  It's almost possible, after this, to scan macros in a

> worker thread.  (However, dwz file handling, and the recent move to

> sharing a bcache, interfere with this.)

> 

> Some smaller utility functions are also moved into other files as

> well.

> 

> Let me know what you think,

> Tom

> 

> 


Other than the little nits I pointed out, this all looks good to me,
thanks for doing this!

Simon
Tom Tromey March 26, 2020, 3:32 p.m. | #2
>>>>> "Simon" == Simon Marchi <simark@simark.ca> writes:


Simon> Other than the little nits I pointed out, this all looks good to me,
Simon> thanks for doing this!

Thanks.  I fixed all of those, and I re-regression-tested the series on
x86-64 Fedora 28.  I'm checking it in now.

Tom