From: Simon Marchi <simon.marchi@efficios.com>
To: Pedro Alves <pedro@palves.net>, gdb-patches@sourceware.org
Subject: Re: [PATCH 00/24] C++ification of struct so_list
Date: Tue, 17 Oct 2023 15:53:25 -0400 [thread overview]
Message-ID: <e6af66d3-e587-4d06-84a0-64e8e71bc148@efficios.com> (raw)
In-Reply-To: <b11ea186-61f0-42e0-a15c-0d1a3bd2e2f3@palves.net>
On 10/17/23 15:20, Pedro Alves wrote:
> On 2023-10-10 21:39, Simon Marchi wrote:
>> This series modernizes the struct so_list area by C++ifying it a little
>> bit and replacing the manual linked list implementation with
>> intrusive_list. It also contains a few other cleanups written along the
>> way.
>
> Very nicely split. Thanks. I sent a few comments, but nothing too serious.
> The only thing on my end that requires some churn is the pointer vs reference
> thing in the disposer. Otherwise, all LGTM.
I dropped the "use reference in disposer" patch.
> Oh, I did notice that "struct so" ends up as an ungreppable/unsearchable
> name, and wondered whether that is going to make our lives harder in practice.
> Maybe we should consider renaming it to "shared_obj" or "shobj" or something easier
> to grep. Or maybe it doesn't matter. But no real strong opinion, I think I'm OK
> with "so" too.
I wondered about that too. If we are going to change the name, perhaps
we can find a more accurate one. I'm currently working on trying to
make it possible for program spaces to have more than one solib
implementation providing so_lists / sos. In the context of ROCm, this
will allow solib-svr4 to provide sos for the host, and solib-rocm to
provide sos for the GPU. I will also try to see if the JIT mechanism
can then be changed to be one more solib provider, see if that
simplifies some things. In the case of ROCm, we don't talk about shared
objects, but code objects that are loaded in the memory space. In the
case of JIT... not sure what we call the generated code, but it's not
really a "shared object". Perhaps there is a generic name we can find
for those pieces of code that can be loaded and unloaded at runtime.
In the mean time, I'll go with "so", since it just seems weird to have
intrusive_list<so_list>. If we want to change it again it's not really
complicated, there aren't that many instances.
Simon
next prev parent reply other threads:[~2023-10-17 19:53 UTC|newest]
Thread overview: 56+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-10 20:39 Simon Marchi
2023-10-10 20:39 ` [PATCH 01/24] gdb: remove empty clear_solib functions Simon Marchi
2023-10-10 20:39 ` [PATCH 02/24] gdb: add program_space parameter to target_so_ops::clear_solib Simon Marchi
2023-10-17 14:57 ` Pedro Alves
2023-10-17 15:19 ` Simon Marchi
2023-10-10 20:39 ` [PATCH 03/24] gdb: make interps_notify work with references Simon Marchi
2023-10-11 8:48 ` Lancelot SIX
2023-10-11 14:18 ` Simon Marchi
2023-10-10 20:39 ` [PATCH 04/24] gdb: replace some so_list parameters to use references Simon Marchi
2023-10-19 11:07 ` [PATCH 4/24] " Lancelot SIX
2023-10-19 14:49 ` Simon Marchi
2023-10-19 15:20 ` Lancelot SIX
2023-10-19 16:07 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 05/24] gdbsupport: use "reference" and "pointer" type aliases in intrusive_list Simon Marchi
2023-10-10 20:40 ` [PATCH 06/24] gdbsupport: make intrusive_list's disposer accept a reference Simon Marchi
2023-10-12 19:05 ` Pedro Alves
2023-10-14 20:12 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 07/24] gdb: make get_cbfd_soname_build_id static Simon Marchi
2023-10-10 20:40 ` [PATCH 08/24] gdb: allocate so_list with new, deallocate with delete Simon Marchi
2023-10-10 20:40 ` [PATCH 09/24] gdb: rename lm_info_base to lm_info Simon Marchi
2023-10-10 20:40 ` [PATCH 10/24] gdb: remove target_so_ops::free_so Simon Marchi
2023-10-10 20:40 ` [PATCH 11/24] gdb: use gdb::checked_static_cast when casting lm_info Simon Marchi
2023-10-10 20:40 ` [PATCH 12/24] gdb: make solib-svr4 not use so_list internally Simon Marchi
2023-10-13 17:52 ` Lancelot SIX
2023-10-14 19:59 ` Simon Marchi
2023-10-19 11:08 ` Lancelot SIX
2023-10-19 14:50 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 13/24] gdb: make solib-rocm " Simon Marchi
2023-10-13 18:35 ` Lancelot SIX
2023-10-14 20:00 ` Simon Marchi
2023-10-17 15:23 ` Pedro Alves
2023-10-17 15:32 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 14/24] gdb: remove lm_info_vector typedef Simon Marchi
2023-10-10 20:40 ` [PATCH 15/24] gdb: make so_list::lm_info a unique_ptr Simon Marchi
2023-10-10 20:40 ` [PATCH 16/24] gdb: make clear_so a method of struct so_list Simon Marchi
2023-10-19 11:08 ` Lancelot SIX
2023-10-19 14:52 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 17/24] gdb: remove target_section_table typedef Simon Marchi
2023-10-10 20:40 ` [PATCH 18/24] gdb: make so_list::sections not a pointer Simon Marchi
2023-10-10 20:40 ` [PATCH 19/24] gdb: make so_list::abfd a gdb_bfd_ref_ptr Simon Marchi
2023-10-10 20:40 ` [PATCH 20/24] gdb: make so_list::{so_original_name,so_name} std::strings Simon Marchi
2023-10-13 22:28 ` [PATCH 20/24] gdb: make so_list::{so_original_name, so_name} std::strings Lancelot SIX
2023-10-14 20:01 ` Simon Marchi
2023-10-19 11:08 ` [PATCH 20/24] gdb: make so_list::{so_original_name,so_name} std::strings Lancelot SIX
2023-10-19 14:55 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 21/24] gdb: link so_list using intrusive_list Simon Marchi
2023-10-17 19:14 ` Pedro Alves
2023-10-17 19:38 ` Simon Marchi
2023-10-10 20:40 ` [PATCH 22/24] gdb: don't call so_list::clear in free_so Simon Marchi
2023-10-10 20:40 ` [PATCH 23/24] gdb: remove free_so function Simon Marchi
2023-10-10 20:49 ` [PATCH 24/24] gdb: rename struct so_list to so Simon Marchi
2023-10-17 19:20 ` [PATCH 00/24] C++ification of struct so_list Pedro Alves
2023-10-17 19:53 ` Simon Marchi [this message]
2023-10-20 14:40 ` Pedro Alves
2023-10-19 11:09 ` [PATCH 0/24] " Lancelot SIX
2023-10-19 14:57 ` Simon Marchi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e6af66d3-e587-4d06-84a0-64e8e71bc148@efficios.com \
--to=simon.marchi@efficios.com \
--cc=gdb-patches@sourceware.org \
--cc=pedro@palves.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).