public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: stsp <stsp2@yandex.ru>
To: Carlos O'Donell <carlos@redhat.com>, libc-alpha@sourceware.org
Subject: Re: [PATCH v9 0/13] implement dlmem() function
Date: Wed, 29 Mar 2023 22:03:34 +0500	[thread overview]
Message-ID: <7fab954f-4a3d-0df0-8211-a9697f3966c1@yandex.ru> (raw)
In-Reply-To: <481ef2c5-a59f-dea4-7f5f-2fcd229a4c25@redhat.com>

[-- Attachment #1: Type: text/plain, Size: 861 bytes --]


29.03.2023 18:17, Carlos O'Donell пишет:
> On 3/18/23 12:50, Stas Sergeev via Libc-alpha wrote:
>
> A cover letter needs to explain in detail what the series does and why glibc
> should include the series e.g. use cases, workloads.
Attaching a cover letter for the upcoming
v10. The patch-set is not yet completed so
please let me know if the cover is now
adequate or needs more descriptions.

Also it would be good if we settle on this
"elf header" stuff before v10 is released,
so I am eagerly waiting your reply on whether
the proof was enough or not.
I can try to bring you any proof you want,
but I can't be asked in a way "its possible
to create an elf that can break your impl,
please prove the opposite". Proving that
something can't be created, usually requires
infinite amount of time. So any proof you
want, needs to be finite and well-defined.

[-- Attachment #2: 0000-cover-letter.patch --]
[-- Type: text/x-patch, Size: 3903 bytes --]

From 46e5095ebfe63be4dcd813c4237d6a491a3f9768 Mon Sep 17 00:00:00 2001
From: Stas Sergeev <stsp2@yandex.ru>
Date: Mon, 13 Feb 2023 18:15:34 +0500
Subject: [PATCH v10 0/12] implement dlmem() function

Changes in v10:
- squashed patch 1 as suggested by Adhemerval Zanella
- fixed a few bugs in an elf relocation machinery after various hot discussions
- added a new test tst-dlmem-dloff that demo-implements dlopen_with_offset()

Changes in v9:
- use "zero-copy" machinery instead of memcpy(). It works on linux 5.13
  and newer, falling back to memcpy() otherwise. Suggested by Florian Weimer.
- implement fdlopen() using the above functionality. It is in a new test
  tst-dlmem-fdlopen. Suggested by Carlos O'Donell.
- add DLMEM_DONTREPLACE flag that doesn't replace the backing-store mapping.
  It switches back to memcpy(). Test-case is called tst-dlmem-shm.

Changes in v8:
- drop audit machinery and instead add an extra arg (optional pointer
  to a struct) to dlmem() itself that allows to install a custom premap
  callback or to specify nsid. Audit machinery was meant to allow
  controling over the pre-existing APIs like dlopen(), but if someone
  ever needs such extensions to dlopen(), he can trivially implement
  dlopen() on top of dlmem().

Changes in v7:
- add _dl_audit_premap audit extension and its usage example

Changes in v6:
- use __strdup("") for l_name as suggested by Andreas Schwab

Changes in v5:
- added _dl_audit_premap_dlmem audit extension for dlmem
- added tst-auditmod-dlmem.c test-case that feeds shm fd to dlmem()

Changes in v4:
- re-target to GLIBC_2.38
- add tst-auditdlmem.c test-case to test auditing
- drop length page-aligning in tst-dlmem: mmap() aligns length on its own
- bugfix: in do_mmapcpy() allow mmaps past end of buffer

Changes in v3:
- Changed prototype of dlmem() (and all the internal machinery) to
  use "const unsigned char *buffer" instead of "const char *buffer".

Changes in v2:
- use <support/test-driver.c> instead of "../test-skeleton.c"
- re-target to GLIBC_2.37
- update all libc.abilist files

This patch-set implements the dlmem() function that allows to load
the solib from page-aligned memory buffer. It has lots of optional
functionality for the fine-grained control over the loading process.
The API looks as below:

/* Callback for dlmem. */
typedef void *
(dlmem_premap_t) (void *mappref, size_t maplength, size_t mapalign,
	          void *cookie);

/* Do not replace mapping created by premap callback.
   dlmem() will then use memcpy(). */
#define DLMEM_DONTREPLACE 1

struct dlmem_args {
  /* Optional name to associate with the loaded object. */
  const char *soname;
  /* Namespace where to load the object. */
  Lmid_t nsid;
  /* dlmem-specific flags. */
  unsigned flags;
  /* Optional premap callback. */
  dlmem_premap_t *premap;
  /* Optional argument for premap callback. */
  void *cookie;
};

/* Like `dlmopen', but loads shared object from memory buffer.  */
extern void *dlmem (const unsigned char *buffer, size_t size, int mode,
		    struct dlmem_args *dlm_args);


In most cases dlm_args should just be set to NULL. It provides the
advanced functionality, most of which is obvious (soname, nsid).
The premap callback allows to set the relocation address for the solib.
More so, if DLMEM_DONTREPLACE flag is used, then the mapping established
by the premap callback, will not be replaced with the file-backed mapping.
In that case dlmem() have to use memcpy(), which is likely even faster
than mmaps() but doesn't end up with the proper /proc/self/map_files
or /proc/self/maps entries. So for example if the premap callback uses
MAP_SHARED, then with the use of the DLMEM_DONTREPLACE flag you can get
your solib relocated into a shared memory buffer. Such functionality
may be interesting for virtualized environments where the relocation
address may have a special constraints, like eg MAP_32BIT.

-- 
2.37.2


  parent reply	other threads:[~2023-03-29 17:03 UTC|newest]

Thread overview: 107+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-03-18 16:50 Stas Sergeev
2023-03-18 16:50 ` [PATCH 01/13] elf: strdup() l_name if no realname [BZ #30100] Stas Sergeev
2023-03-29 13:54   ` Adhemerval Zanella Netto
2023-03-29 14:12     ` stsp
2023-03-29 14:19       ` Adhemerval Zanella Netto
2023-03-29 14:28         ` stsp
2023-03-29 14:30           ` Adhemerval Zanella Netto
2023-03-29 14:33             ` stsp
2023-03-18 16:50 ` [PATCH 02/13] elf: switch _dl_map_segment() to anonymous mapping Stas Sergeev
2023-03-29 17:01   ` Adhemerval Zanella Netto
2023-03-29 18:00     ` stsp
2023-03-29 18:29       ` Adhemerval Zanella Netto
2023-03-29 18:46         ` stsp
2023-03-29 19:17           ` Adhemerval Zanella Netto
2023-03-29 19:43             ` stsp
2023-03-18 16:51 ` [PATCH 03/13] elf: dont pass fd to _dl_process_pt_xx Stas Sergeev
2023-03-29 17:10   ` Adhemerval Zanella Netto
2023-03-30 16:08     ` stsp
2023-03-30 20:46       ` Adhemerval Zanella Netto
2023-03-31 12:02         ` Szabolcs Nagy
2023-03-31 12:54           ` Adhemerval Zanella Netto
2023-03-31 14:04             ` stsp
2023-03-18 16:51 ` [PATCH 04/13] elf: split _dl_map_object_from_fd() into reusable parts Stas Sergeev
2023-03-18 16:51 ` [PATCH 05/13] elf: split open_verify() " Stas Sergeev
2023-03-18 16:51 ` [PATCH 06/13] elf: load elf hdr fully in open_verify() Stas Sergeev
2023-03-18 16:51 ` [PATCH 07/13] elf: convert pread64 to callback in do_open_verify() Stas Sergeev
2023-03-18 16:51 ` [PATCH 08/13] elf: convert _dl_map_segments's mmap() to a callback Stas Sergeev
2023-03-18 16:51 ` [PATCH 09/13] elf: call _dl_map_segment() via premap callback Stas Sergeev
2023-03-18 16:51 ` [PATCH 10/13] elf: convert _dl_map_object to a callback Stas Sergeev
2023-03-18 16:51 ` [PATCH 11/13] elf: split _dl_check_loaded() from _dl_map_object Stas Sergeev
2023-03-18 16:51 ` [PATCH 12/13] dlfcn,elf: implement dlmem() [BZ #11767] Stas Sergeev
2023-03-29 13:45   ` Carlos O'Donell
2023-03-29 13:51     ` stsp
2023-03-29 14:10       ` Jonathon Anderson
2023-03-29 14:20         ` stsp
2023-03-29 14:31           ` Adhemerval Zanella Netto
2023-03-29 15:01             ` stsp
2023-03-29 14:35           ` Carlos O'Donell
2023-03-29 14:50             ` stsp
2023-03-29 15:20               ` Carlos O'Donell
2023-03-29 15:34                 ` stsp
2023-03-30  8:09         ` stsp
2023-03-18 16:51 ` [PATCH 13/13] dlfcn,elf: impl DLMEM_DONTREPLACE dlmem() flag Stas Sergeev
2023-03-29 12:32 ` [PATCH v9 0/13] implement dlmem() function Adhemerval Zanella Netto
2023-03-29 13:10   ` stsp
2023-03-29 13:18   ` stsp
2023-03-31 12:20     ` Szabolcs Nagy
2023-03-31 13:51       ` stsp
2023-03-31 14:49         ` Rich Felker
2023-03-31 14:56           ` stsp
2023-03-31 14:58             ` Rich Felker
2023-03-31 15:03               ` stsp
2023-03-31 14:44       ` stsp
2023-03-31 15:12       ` stsp
2023-03-31 17:12         ` Szabolcs Nagy
2023-03-31 17:36           ` stsp
2023-04-01  9:28             ` stsp
2023-04-03 10:04             ` Szabolcs Nagy
2023-04-03 10:43               ` stsp
2023-04-03 12:01                 ` Szabolcs Nagy
2023-04-03 13:07                   ` stsp
2023-04-05  7:29                   ` stsp
2023-04-05  8:51                     ` Szabolcs Nagy
2023-04-05  9:26                       ` stsp
2023-04-05  9:31                       ` Florian Weimer
2023-04-12 17:23                       ` stsp
2023-04-12 18:00                         ` stsp
2023-04-12 18:20                           ` Rich Felker
2023-04-12 18:46                             ` stsp
2023-04-12 19:52                               ` Zack Weinberg
2023-04-12 19:07                             ` stsp
2023-04-13 10:01                             ` stsp
2023-04-13 12:38                               ` Szabolcs Nagy
2023-04-13 15:59                                 ` stsp
2023-04-13 18:09                                   ` Adhemerval Zanella Netto
2023-04-13 18:59                                     ` stsp
2023-04-13 19:12                                       ` Adhemerval Zanella Netto
2023-04-13 19:29                                         ` stsp
2023-04-13 20:02                                           ` Adhemerval Zanella Netto
2023-04-13 20:21                                             ` stsp
2023-04-13 20:57                                             ` stsp
2023-04-14  7:07                                             ` stsp
2023-04-14  7:36                                             ` stsp
2023-04-14 11:30                                             ` stsp
2023-04-14 19:04                                             ` proof for dlmem() (Re: [PATCH v9 0/13] implement dlmem() function) stsp
2023-05-01 23:11                                               ` Zack Weinberg
2023-05-02  5:48                                                 ` stsp
2023-05-08 16:00                                                   ` stsp
2023-05-02  6:24                                                 ` stsp
2023-05-08 15:10                                 ` [PATCH v9 0/13] implement dlmem() function stsp
2023-03-31 18:47           ` stsp
2023-03-31 19:00             ` stsp
2023-03-29 13:17 ` Carlos O'Donell
2023-03-29 13:26   ` stsp
2023-03-29 17:03   ` stsp [this message]
2023-03-29 18:13     ` Carlos O'Donell
2023-03-29 18:29       ` stsp
2023-03-31 11:04       ` stsp
2023-04-13 21:17         ` Carlos O'Donell
2023-04-13 21:58           ` stsp
2023-04-13 22:08           ` stsp
2023-04-13 22:50           ` stsp
2023-04-14 16:15           ` Autoconf maintenance (extremely tangential to Re: [PATCH v9 0/13] implement dlmem() function) Zack Weinberg
2023-04-14 20:24             ` Carlos O'Donell
2023-04-14 20:40               ` Zack Weinberg
2023-05-08 15:05           ` [PATCH v9 0/13] implement dlmem() function stsp
2023-05-19  7:26           ` stsp

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=7fab954f-4a3d-0df0-8211-a9697f3966c1@yandex.ru \
    --to=stsp2@yandex.ru \
    --cc=carlos@redhat.com \
    --cc=libc-alpha@sourceware.org \
    /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).