March 2001 - Wine-Devel - list.winehq.org

Re: Speeding up wineserver syncronization objects with shared memory
by Alexandre Julliard March 8, 2001

March 8, 2001

David Howells <dhowells(a)cambridge.redhat.com> writes: > Not so... The file offset must also be page aligned! OK, this has apparently changed in kernel 2.4; in kernel 2.2 block size alignment is enough. That's unfortunate, but I guess it was necessary to support large files or memory sizes. Then yes, in 2.4 your patch will be more useful. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Speeding up wineserver syncronization objects with shared memory
by Alexandre Julliard March 8, 2001

March 8, 2001

David Howells <dhowells(a)cambridge.redhat.com> writes: > My main gripe is the slow speed of access to files... Every Read/WriteFile > goes to the wineserver to convert the handle into a file descriptor and to > check for locking. The FD is then passed back over a UNIX domain socket, used > once and then closed. Note that we are no longer doing that in the latest versions; the file descriptor is only transferred once, and all further requests are done on a pipe which is faster than a socket. > Gavriel State <gav(a)transgaming.com> wrote: > > In my experience Alexandre far prefers incremental change to kind of > > approach you're taking. Using that kind of approach will improve the > > chances that your code will make it into Wine at some point. > > Hmmm... It's difficult to determine how to do it incrementally without making > for even more work, but I think I know what you mean. The kernel module itself may be hard to do incrementally, but you should really consider reusing the existing server API so that your module can be plugged in easily. For instance your module entry points should be the same as the server requests, and use the same request structures. > Context switching is the main element of it. Going to the wineserver and back > again just for a ReadFile() call or a Wait*() function incurs a fairly serious > penalty (particularly on an X86, I think). Plus there's no requirement for the > kernel to pass the remains of your timeslice to the wineserver and back > again. I still think that it should be possible to improve that by a small kernel hack. It will never be as fast as doing everything in the kernel of course, but it may just be fast enough to avoid the need to reimplement the whole server. > Look at the VM size of something like MS Word, and think of not having to > allocate buffers to store all those DLLs, thus eating a massive chunk out of > your machines total VM. Have you measured how many dirty pages you can avoid with your change? It seems to me that in most cases, when the dll is loaded at its preferred address, the number of pages made dirty by the fixups should be quite small anyway. -- Alexandre Julliard julliard(a)winehq.com

2 2

Re: Speeding up wineserver syncronization objects with shared memory
by Alexandre Julliard March 8, 2001

March 8, 2001

Gavriel State <gav(a)transgaming.com> writes: > Looking at the kernel sources, it appears to be page-alignment, not > filesystem block size that mmap needs to match. Rereading old wine-devel > archives, it appears that the notion that the filesystem block size is > involved is either from a very old kernel, or a misreading of some kernel > comments. Page alignment is needed for the address in memory, not for the offset inside the file on disk; since section virtual addresses in PE files are always page-aligned the memory address is never a problem. The only problem comes from the alignment of the data inside the PE file, and this is where we only need block-size alignment to make mmap possible. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Speeding up wineserver syncronization objects with shared memory
by Alexandre Julliard March 7, 2001

March 7, 2001

David Howells <dhowells(a)cambridge.redhat.com> writes: > So this saves you the cost of the fd transfer net packet. Though you still > have to do the two context switches, which is my main contention. I suspect we are doing more than two switches (though I haven't proved it), which is why I think there is a margin for improvement. You'll obviously always have the context switch cost unless everything is in the kernel. > True, but I'd have thought that the context switches involved are still a cost > you can't get rid of so easily. Out of interest, how do you plan on doing the > locking stuff for Read/WriteFile? Cache it locally? It is unfortunate, but you > can't really make use of UNIX file locking, since this is mostly advisory and > as such doesn't actively stop read/write calls. Yes, we'll need to store the locks in the server and check them before each read/write (and probably also release them afterwards if necessary). There may be some optimisations possible, but we should probably do it the easy way first. > Seriously, though, whilst this'd be a lot easier in many ways (and it would > allow you to avoid the context-switch penalties), you wouldn't be able to take > full advantage of the available support in the kernel, which is more capable > than the standard UNIX userspace API suggests. I don't see why. I'm not suggesting keeping the current socket stuff, just reusing the structures. So basically instead of passing the address of the stack arguments (which is really ugly IMO) to your ioctl, you pass one of the server request structures. This allows your changes to be localized to wine_server_call and doesn't require changing any of the routines that make server calls. Obviously you'd need some more changes for a few calls like ReadFile/WriteFile, but most operations could switch to your mechanism without needing any change. You simply cannot require people to recompile all of Wine to use your module. > > I still think that it should be possible to improve that by a small > > kernel hack. It will never be as fast as doing everything in the > > kernel of course, but it may just be fast enough to avoid the need to > > reimplement the whole server. > > If you want to suggest exactly what you'd like to see as a hack... I don't know exactly, there are many ways of doing it; you can have a specialized fifo, a network protocol, an ioctl, etc. Basically any mechanism that ensures that we do the strict mimimum number of context switches and schedule() calls for a server call. And probably also a way to transfer chunks of memory from the client address space so that we don't need the shared memory area. > As far as I've observed (I've got Win2000 available), most Windows DLL's have > 512-byte (sector) alignment internally, _not_ 4096-byte (page) alignment for > the sections. This means that the separate sections can't be mmap'd (or else > they'd lose their required relative relationships): Actually the file alignment doesn't need to be 4096, it needs to match the filesystem block size. On a FAT filesystem the block size is 512 so Linux will happily mmap every section. On a 1k-block ext2 fs it will be able to mmap about 50% of them. > Also, since DLLs and EXEs are not compiled as PIC (the MSDEV compiler not > having such an option as far as I can recall), the fixup tables usually seem > to apply to just about every page in the code section. Only if the dll cannot be loaded at the preferred address, which shouldn't happen too often. I'm not saying your patch is useless, but I doubt the gain is as large as you seem to think. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Wineserver semantics
by Alexandre Julliard March 6, 2001

March 6, 2001

Ryan Cumming <bodnar42(a)bodnar42.dhs.org> writes: > What about having the file descriptor cache explictly invaildated? Like, > the server keeps track of which file descriptors the various wine > processes have cached, and if a process closes or performs some locking > operation on a file descriptor, all the other wine proceses that are using > that file descriptor are explictly notified. This way processes can assume > a file descriptor is valid unless told otherwise... would this be any > better performance-wise? Does it fit in to the wineserver architecture > nicely? ;) Not really; you'd either need to callback into the client, which creates a lot of synchronisation issues, or use a shared memory area which implies locking. Either way doesn't really fit in the current model. And as mentioned already we'll probably still need a server call on each file access to implement file locking. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Wineserver semantics
by Alexandre Julliard March 6, 2001

March 6, 2001

Waldek Hebisch <hebisch(a)math.uni.wroc.pl> writes: > There is ongoing disscusion on how to seed up the wineserver. > So I have question concerning correctness of corrent implementation. > Namely, for reading or writing files I have not found any locking in > wineserver when issuing file descriptor to clients. On the other > hand the client closes file descriptor just after use, and also > disscusion in the group suggests that client is supposed to have > exclusive acces to file. So I would like to know what is the case: > 1) client is allowd to cache the descriptor > 2) current implementation depends on winning a race > 3) I missed locking in wineserver I'm not sure what you mean by 'exclusive access to the file'. If you mean file locking then this is not implemented, so there cannot be any race ;-) Currently the client is allowed to cache the descriptor, but before each read/write it checks with the server that the cache is still valid. There is a race that the handle can be modified between the check and the actual read, this will cause either a bad handle error or a read from the wrong file; but this is no different from what happens under Windows if the client is manipulating handles concurrently with a ReadFile. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: winelauncher patch
by Jeremy White March 6, 2001

March 6, 2001

Ove, I have in fact determined why I did not use which: On Suse (and Caldera, I believe), which /usr/bin/wine returns an error string, making the whole script go straight down the drain. What was wrong with the previous implemenation? Are you getting a $0 value that doesn't include the path? In all of the test cases I have, invoking wine seems to bring in a $0 that includes enough of the path for the find logic to work. Jer > Hunk 2: > real_name=`find $0` don't find anything if the user typed "wine" or > something, wine is in /usr/bin, which is in PATH, but the cwd is ~... > rather run it through `which $0` first. Also, the symlink it's reading > don't have to be absolute (my symlinks are relative), deal with that. > > -real_name=`find $0 -type l -printf "%l\n"` > +argv0_path=`which $0` > +if [ -z $argv0_path ] ; then > + argv0_path=$0 > +fi > + > +real_name=`find $argv0_path -type l -printf "%l\n"` > if [ ! $real_name ]; then > - real_name=$0; > + real_name=$argv0_path > +elif [ ! -x $real_name ]; then > + real_name=`find $argv0_path -printf "%h\n"`/$real_name > fi > + > argv0_dir=`find $real_name -printf "%h\n"` > > if [ -z $argv0_dir ] ; then > @@ -130,7 +138,7 @@ > WINESERVER=$WINEBIN/server/wineserver > fi

2 1

Wineserver semantics
by Waldek Hebisch March 6, 2001

March 6, 2001

There is ongoing disscusion on how to seed up the wineserver. So I have question concerning correctness of corrent implementation. Namely, for reading or writing files I have not found any locking in wineserver when issuing file descriptor to clients. On the other hand the client closes file descriptor just after use, and also disscusion in the group suggests that client is supposed to have exclusive acces to file. So I would like to know what is the case: 1) client is allowd to cache the descriptor 2) current implementation depends on winning a race 3) I missed locking in wineserver -- Waldek Hebisch hebisch(a)math.uni.wroc.pl or hebisch(a)hera.math.uni.wroc.pl

1 0

WinCE/PocketWindows?
by Robert & Jennifer Hartley March 6, 2001

March 6, 2001

Hi, (Please forgive me if this has already been examined, as this is my first post.) I am curious to know if a version of WINE has been considered that would be made to run WinCE (CE) and PocketWindows (PW) applications on Linux PDAs. I have been working with the Agenda VR3 PDA from www.agendacomputing that runs Linux very well on a MIPS processor. I also received a free set of Embedded Tools CDs from MicroSoft that have VC++ and VB on them for building CE and PW applications. The platforms it supports are X86, MIPS, ARM, and SH3, and I believe there are others. There is a compatability layer that runs on WinNT to test before deploying. If we had that emulated in Linux, then we would be most of the way towards being done. (Or so I would imagine.) Naturally, since I am a Linux/Unix guy, I was thinking that it would be neat to have these CE/PW apps running on my X86 Linux box, my NetWinder, my venerable old Indy, and finally, once debugged, on this Agenda Linux PDA. The WINE community has already done a huge amount of great work, so I thought that maybe a cut down version of WINE would work on the CE/PW subsets of Windows. Being an outsider, could someone let me know if I am totally out there in LaLa Land? Thanks, Rob (here around Ottawa, ON) -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= = Robert Hartley Mail: 201 Broadway = = Central Region Systems Engineer Cambridge, MA 02139 = = Integrated Computer Email: rhartley(a)ics.com = = Solutions, Inc. Web Site: www.ics.com = = Tech Support: support(a)ics.com Phone: 800-800-4271 = =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Visit the MotifZone (www.motifzone.org) for info on Motif!

1 0

Re: Perl/Wine extension for perusal
by Niki Spahiev March 6, 2001

March 6, 2001

Hello, Two small suggestions: 1. python has very good WIN32 bindings and also generic DLL interface CALLDLL. 2. VNC has alpha of macro recorder which can be used for interactive tests. Regards, Niki Spahiev

1 0