Here is a solution for this problem. Please, review.
To reproduce the crash write manually a path with non-existent drive
in the local package directory edit control.
The reason for this crash is that mkdir_p tries to create the path
recursevly and does not care if its actually a path component
new return value is a problem for
any of the mkdir_p () callers.
Here is the ChangeLog entry:
2002-03-26 Pavel Tsekov <[EMAIL PROTECTED]>
* mkdir.cc (mkdir_p): Stop processing if the path is exhausted.
Return -1 on 'path exhausted' condition.
>> Btw. I haven't checked if the new return value is a problem
>> for any of the mkdir_p () callers.
RC> Oh, it may well be, so I was going to checkin a variant anyway:
RC> 1 is failure :}.
Uh, yes... Thanks! I was sure I'm missing something :)
Hello, there! :)
Wednesday, April 10, 2002, 10:11:46 AM, you wrote:
Ok, found it - its canonicalize_version() in version.cc. There is
a pointer 'v' which is modified and then delete[]'d.
A trivial patch is attached :)
2002-04-10 Pavel Tsekov <[EMAIL PROTECTED]>
No comments on this at all ?
PT> Ok, found it - its canonicalize_version() in version.cc. There is
PT> a pointer 'v' which is modified and then delete[]'d.
PT> A trivial patch is attached :)
PT> 2002-04-10 Pavel Tsekov <[EMAIL PROTECTED]>
PT>
Please, ignore this - I was too speedy to type :( Just saw that a new
version was released
Friday, April 12, 2002, 10:02:37 AM, you wrote:
PT> No comments on this at all ?
Hello,
While I was browsing the fhandler_registry code and the *stat()
stuff it occured to me that this patch may be useful. I don't know
how it improves the speed of the path_conv::check (I don't think it
is a noticable improvement), but for sure it saves some cpu cycles.
2002-06
Hello,
Currently "rm -rf" on a file or directory under /proc returns
ENOENT. This is not correct and is caused by the fact that
the posix filename for, say, /proc/uptime is translated to
the following win32 name: c:\cygwin\proc\uptime (if c:\cygwin
is mounted on /).
The attached patches fix this
The attached dir.cc.diff in the original mail was wrong. Please, find attached the
proper
fix for dir.cc.
Sorry for the inconvenience :(
PT> The attached patches fix this problem by returning EROFS
dir.cc.diff
Description: Binary data
They are very small:
syscalls.cc.diff and dir.cc.diff have 8 lines added.
path.cc.diff has line one changed.
Does this require an assignment ?
CF> I haven't actually read your patch since, AFAIK, you don't have an
CF> assignment with Red Hat, right? If this is the case, would you mind
CF> send
Hello Christopher,
Tuesday, June 04, 2002, 5:35:35 PM, you wrote:
CF> I haven't actually read your patch since, AFAIK, you don't have an
CF> assignment with Red Hat, right? If this is the case, would you mind
CF> sending one in?
Yes, I don't have an assignment. Today I've started the process t
I see that this patch is no longer required. The snapshot from July
4th includes a similiar change.
PT> 2002-06-04 Pavel Tsekov <[EMAIL PROTECTED]>
PT> * path.cc (path_conv::check): Do not check a directory
PT> path against a known list of executable file extensions.
CF> Btw, make sure that your employer signs their part, too.
It's been aranged :) I wouldn't even think of asking if it wasn't.
I'm just waiting for the disclaimer to be printed on a paper with the
company sign on it :)
On Wed, 26 Mar 2003, Corinna Vinschen wrote:
> However... am I doing something wrong? I'm trying to find out what the
> performance improvement is on my XP box and both versions of the DLL
> (w/ and w/o your patch) are running 7.5 minutes for
>
> ls -lR /proc/registry > /dev/null
>
> Or is t
Hello,
Does the following patch make any sense ?
* 2003-07-21 Pavel Tsekov <[EMAIL PROTECTED]>
* mmap.cc: Use proper format specifiers for _off64_t and size_t in
format strings passed to syscall_printf () and debug_printf ()
throughout.
--
+++ GMX - Mail, Mes
ative just in case. The attached
patch is pretty much non-intrusive, while the one employing cygwin_internal ()
adds a new member to the tty_min structure thus changing the shared memory
footprint and I consider it too intrusive.
Please, review and share your thoughts :)
2003-08-01 Pavel Tseko
Hello,
Here is a simple patch which makes the behaviour of getmntent () consistent
with
the one of fhandler_cygdrive.
Pavel
2003-08-02 Pavel Tsekov <[EMAIL PROTECTED]>
path.cc (cygdrive_getmntent): Do not skip over drives of
type DRIVE_REMOVABLE.
--
COMPUTERBILD
> On Sat, Aug 02, 2003 at 01:53:21PM +0200, Pavel Tsekov wrote:
> >Here is a simple patch which makes the behaviour of getmntent ()
> >consistent with the one of fhandler_cygdrive.
>
> The reason this is there is to avoid long delays in mount table
> listings.
Well, I gue
> After giving this a couple days of thought, I decided that this was
> not hobgoblin related so I'm going to check this in.
Thanks! Of course, if people start complaining, it could easily reverted.
Pavel
--
COMPUTERBILD 15/03: Premium-e-mail-Dienste im Test
Hello,
Should the silence indicate a lack of interest ?
Pavel
--
COMPUTERBILD 15/03: Premium-e-mail-Dienste im Test
--
1. GMX TopMail - Platz 1 und Testsieger!
2. GMX ProMail - Platz 2 und Preis-Qualitätssieger!
3. Arcor - 4. web.de - 5. T-Online
Hello,
Attached is a patch + a testcase. The testcase can be executed with any of
the
following directories:
/cygdrive
/proc
/proc/
/proc/registry/HKEY_WHATEVER
2003-08-05 Pavel Tsekov <[EMAIL PROTECTED]>
* fhandler_disk_file.cc (fhandler_cygdrive::readdir): Do not
-03-06 Pavel Tsekov <[EMAIL PROTECTED]>
* path.cc (mount_info::read_cygdrive_info_from_registry):
Use the user prefix if it exists.
Index: path.cc
===
RCS file: /cvs/src/src/winsup/cygwin/path.cc,v
retrieving revisio
Hello,
I've been looking at the contents of the cygwin1.dll image in the last
few days and I've noticed that the sections named .dllname_info contain
a lot of duplicate entries - one for each autoloaded function from a given
dll. Although it doesn't hurt to have it, this information is not really
On Thu, 10 Mar 2005, Christopher Faylor wrote:
> Thanks for the patch but I'd prefer keeping the current functionality which
> makes the use of LoadDllprime optional.
I didn't know better :(
> I've checked in a fix for this so that only one .*_info block is loaded
> for any given DLL.
Very nic
Hello,
Today, I've finally tracked down the cause of the problem with the stale
subshells on MC exit. In the past I've tried to attack the problem in a
different manner and wasn't looking at the right place - I suspected some
kind of problem / race in the signal handling / delivery, but I was wron
On Wed, 17 Aug 2005, Christopher Faylor wrote:
> On Wed, Aug 17, 2005 at 06:30:25PM +0300, Pavel Tsekov wrote:
> >The attached patch solves this problem by rearranging the code a bit. It
> >tries to be non-intrusive. I offer it for discussion and comments. I hope
> >that
26 matches
Mail list logo