* Using pread
@ 2013-02-01 8:05 Dmitry Antipov
2013-02-01 8:18 ` Paul Eggert
0 siblings, 1 reply; 6+ messages in thread
From: Dmitry Antipov @ 2013-02-01 8:05 UTC (permalink / raw)
To: Emacs development discussions; +Cc: Paul Eggert
[-- Attachment #1: Type: text/plain, Size: 173 bytes --]
Can we assume that all supported systems has pread(2)? If yes, we can optimize
read + lseek in some cases (and even if not, it's easy to emulate it in emacs_pread).
Dmitry
[-- Attachment #2: pread.patch --]
[-- Type: text/plain, Size: 2451 bytes --]
=== modified file 'src/fileio.c'
--- src/fileio.c 2013-02-01 06:30:51 +0000
+++ src/fileio.c 2013-02-01 07:40:44 +0000
@@ -3703,17 +3703,14 @@
int nread;
if (st.st_size <= (1024 * 4))
- nread = emacs_read (fd, read_buf, 1024 * 4);
+ nread = emacs_pread (fd, read_buf, 1024 * 4, 0);
else
{
- nread = emacs_read (fd, read_buf, 1024);
+ nread = emacs_pread (fd, read_buf, 1024, 0);
if (nread == 1024)
{
- int ntail;
- if (lseek (fd, - (1024 * 3), SEEK_END) < 0)
- report_file_error ("Setting file position",
- Fcons (orig_filename, Qnil));
- ntail = emacs_read (fd, read_buf + nread, 1024 * 3);
+ int ntail = emacs_pread (fd, read_buf + nread, 1024 * 3,
+ st.st_size - 1024 * 3);
nread = ntail < 0 ? ntail : nread + ntail;
}
}
@@ -3753,11 +3750,6 @@
/* Discard the unwind protect for recovering the
current buffer. */
specpdl_ptr--;
-
- /* Rewind the file for the actual read done later. */
- if (lseek (fd, 0, SEEK_SET) < 0)
- report_file_error ("Setting file position",
- Fcons (orig_filename, Qnil));
}
}
=== modified file 'src/lisp.h'
--- src/lisp.h 2013-02-01 06:30:51 +0000
+++ src/lisp.h 2013-02-01 07:33:29 +0000
@@ -3566,6 +3566,7 @@
extern int emacs_open (const char *, int, int);
extern int emacs_close (int);
extern ptrdiff_t emacs_read (int, char *, ptrdiff_t);
+extern ptrdiff_t emacs_pread (int, char *, ptrdiff_t, ptrdiff_t);
extern ptrdiff_t emacs_write (int, const char *, ptrdiff_t);
extern void unlock_all_files (void);
=== modified file 'src/sysdep.c'
--- src/sysdep.c 2013-02-01 06:30:51 +0000
+++ src/sysdep.c 2013-02-01 07:35:36 +0000
@@ -2209,6 +2209,21 @@
return (rtnval);
}
+/* Read from FILEDESC to a buffer BUF with size NBYTES at offset OFFSET,
+ retrying if interrupted. Return the number of bytes read, which might
+ be less than NBYTES. On error, set errno and return -1. */
+
+ptrdiff_t
+emacs_pread (int fildes, char *buf, ptrdiff_t nbytes, ptrdiff_t offset)
+{
+ register ssize_t rtnval;
+
+ while ((rtnval = pread (fildes, buf, nbytes, offset)) == -1
+ && (errno == EINTR))
+ QUIT;
+ return rtnval;
+}
+
/* Write to FILEDES from a buffer BUF with size NBYTE, retrying if interrupted
or if a partial write occurs. Return the number of bytes written, setting
errno if this is less than NBYTE. */
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Using pread
2013-02-01 8:05 Using pread Dmitry Antipov
@ 2013-02-01 8:18 ` Paul Eggert
2013-02-01 8:34 ` Dmitry Antipov
0 siblings, 1 reply; 6+ messages in thread
From: Paul Eggert @ 2013-02-01 8:18 UTC (permalink / raw)
To: Dmitry Antipov; +Cc: Emacs development discussions
On 02/01/2013 12:05 AM, Dmitry Antipov wrote:
> Can we assume that all supported systems has pread(2)?
A few systems lack pread (HP-UX 10, mingw, MSVC 9, BeOS)
but there's a gnulib module for it so the mainline Emacs
code should be able to assume pread.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Using pread
2013-02-01 8:18 ` Paul Eggert
@ 2013-02-01 8:34 ` Dmitry Antipov
2013-02-01 10:33 ` Eli Zaretskii
2013-02-02 1:47 ` Paul Eggert
0 siblings, 2 replies; 6+ messages in thread
From: Dmitry Antipov @ 2013-02-01 8:34 UTC (permalink / raw)
To: Paul Eggert; +Cc: Emacs development discussions
On 02/01/2013 12:18 PM, Paul Eggert wrote:
> A few systems lack pread (HP-UX 10, mingw, MSVC 9, BeOS)
> but there's a gnulib module for it so the mainline Emacs
> code should be able to assume pread.
IMHO it's worth using at least for some bits in fileio.c.
So if you add gnulib module, I can do the rest :-).
Dmitry
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Using pread
2013-02-01 8:34 ` Dmitry Antipov
@ 2013-02-01 10:33 ` Eli Zaretskii
2013-02-02 1:47 ` Paul Eggert
1 sibling, 0 replies; 6+ messages in thread
From: Eli Zaretskii @ 2013-02-01 10:33 UTC (permalink / raw)
To: Dmitry Antipov; +Cc: eggert, emacs-devel
> Date: Fri, 01 Feb 2013 12:34:27 +0400
> From: Dmitry Antipov <dmantipov@yandex.ru>
> Cc: Emacs development discussions <emacs-devel@gnu.org>
>
> On 02/01/2013 12:18 PM, Paul Eggert wrote:
>
> > A few systems lack pread (HP-UX 10, mingw, MSVC 9, BeOS)
> > but there's a gnulib module for it so the mainline Emacs
> > code should be able to assume pread.
>
> IMHO it's worth using at least for some bits in fileio.c.
Any numbers to back that up? IOW, in what sense is this an
"optimization"?
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Using pread
2013-02-01 8:34 ` Dmitry Antipov
2013-02-01 10:33 ` Eli Zaretskii
@ 2013-02-02 1:47 ` Paul Eggert
2013-02-02 8:09 ` Eli Zaretskii
1 sibling, 1 reply; 6+ messages in thread
From: Paul Eggert @ 2013-02-02 1:47 UTC (permalink / raw)
To: Dmitry Antipov; +Cc: Emacs development discussions
On 02/01/2013 12:34 AM, Dmitry Antipov wrote:
> IMHO it's worth using at least for some bits in fileio.c.
Is using pread significantly faster or safer, when pread is native?
If this is merely a style issue, it's probably not worth the bother.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Using pread
2013-02-02 1:47 ` Paul Eggert
@ 2013-02-02 8:09 ` Eli Zaretskii
0 siblings, 0 replies; 6+ messages in thread
From: Eli Zaretskii @ 2013-02-02 8:09 UTC (permalink / raw)
To: Paul Eggert; +Cc: dmantipov, emacs-devel
> Date: Fri, 01 Feb 2013 17:47:35 -0800
> From: Paul Eggert <eggert@cs.ucla.edu>
> Cc: Emacs development discussions <emacs-devel@gnu.org>
>
> On 02/01/2013 12:34 AM, Dmitry Antipov wrote:
> > IMHO it's worth using at least for some bits in fileio.c.
>
> Is using pread significantly faster or safer, when pread is native?
AFAIU, its only advantage is when several threads might share a file
descriptor. But if we consider this possibility for some future
extension, such as Tom's concurrency branch, then perhaps we should
modify emacs_read to use pread as well.
> If this is merely a style issue, it's probably not worth the bother.
I agree.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2013-02-02 8:09 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-02-01 8:05 Using pread Dmitry Antipov
2013-02-01 8:18 ` Paul Eggert
2013-02-01 8:34 ` Dmitry Antipov
2013-02-01 10:33 ` Eli Zaretskii
2013-02-02 1:47 ` Paul Eggert
2013-02-02 8:09 ` Eli Zaretskii
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/emacs.git
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).