- Wine-devel - list.winehq.org

Re: Nasty Evil Memory Fragmentation fix
by Alexandre Julliard 11 Oct '01

11 Oct '01

Ove Kaaven <ovehk(a)ping.uio.no> writes: > Not that much. Wine's allocator already implements the *features* of that > allocator (boundary tags, bins), it just doesn't implement all the > *strategies* used for selecting which free block to allocate memory from > (though the only big things we lack there is "best fit" and "wilderness > preservation", the latter of which Alexandre talked about). Well, we do have (almost) best fit, that's the only reason for having multiple free lists. It's just that we probably need more bins sizes to make it really useful. > We just need > to make sure Wine's free block selection routine uses the same strategies > as this allocator, and we'd get exactly the same performance as that > allocator, but without replacing the whole thing and sacrificing the > compatibility we have or introducing new bugs. Well, this depends how easy it is to plug the new allocator in; but if it doesn't require too many changes, it will certainly introduce less bugs than trying to tweak the existing code. Plus all the profiling work has been done already, which would have to be done on the Wine code. -- Alexandre Julliard julliard(a)winehq.com

1 0

FIXME:pthread_rwlock_rdlock
by 510045342665-0001＠t-online.de 11 Oct '01

11 Oct '01

Hi, I am running a compiler with wine. There are many fixme message that I can't switch off. What do they mean, is someone working on it, is the problem serious ? Not having them at stdout would be great :-) as the compiler output + wine fixmes get a little confusing... [scan1.c] FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock FIXME:pthread_rwlock_rdlock FIXME:pthread_rwlock_unlock --Dirk

1 0

Re: Nasty Evil Memory Fragmentation fix
by Alexandre Julliard 11 Oct '01

11 Oct '01

Gavriel State <gav(a)transgaming.com> writes: > We ended up creating a new subheap every time one of these larger > blocks came along. The new space from the subheap then went to > the top of the freelist, so all the small allocations then filled > it up until the next large block allocation. IE: We never got a > chance to reuse freed blocks in older subheaps - at least not > very well. That's why we have multiple free lists. The code is using the Win95 values now, but there is no reason we couldn't add a few larger values in the HEAP_freeListSizes array, instead of creating a new mechanism. > Despite the workaround, we're still not too pleased with the current > heap allocator. It's quite slow, and still not as efficient as > it could be. It would probably be worth the effort to integrate > a new allocator - anyone know if there's a high-quality Wine-license- > compatible allocator out there? A new allocator would definitely be a good thing. This one was done the way it is in order to use a compatible memory layout, but it turns out this is not necessary so we could scrap it. -- Alexandre Julliard julliard(a)winehq.com

3 3

Re: fixed FILE_GetUnixHanlde for sockets
by Andreas Mohr 11 Oct '01

11 Oct '01

On Thu, Oct 11, 2001 at 06:53:33PM +0200, Kai Morich wrote: > added sock_get_info in server/sock.c (patch includes changelog) "Patch includes changelog". hmm, patch. which patch ? which changelog ? ;-) -- Andreas Mohr Stauferstr. 6, D-71272 Renningen, Germany Tel. +49 7159 800604 http://home.nexgo.de/andi.mohr/

1 0

SHGetSpecialFolderPathA and CSIDL_FOLDER_MASK
by Chris Green 11 Oct '01

11 Oct '01

I've been looking at this function, and I am starting to suspect we have the wrong definition of CSIDL_FOLDER_MASK (currently 0x00ff). What I've been seeing (eg. installing MS Office 97) seems to imply that only the lower 6 bits, rather than 8, are related to the folder, and the others are something else. Does anyone else think that the mask should be 0x003f? regards Chris _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com

2 1

Re: Nasty Evil Memory Fragmentation fix
by Alexandre Julliard 11 Oct '01

11 Oct '01

Gavriel State <gav(a)transgaming.com> writes: > This still doesn't help us: as Ove points out, we still end up allocating > from newly created free blocks even if old blocks would suit us better. But you want to use new blocks in general. The exception is that you want to avoid using the uncommitted part at the end of the heap, and I agree we could do a special case for that. But it has nothing to do with the size of the block. > I've had a brief look at the code referenced by Adam Moss (found at > http://g.oswego.edu/dl/html/malloc.html) and it looks like it would > be a vast improvement over Wine's current Heap manager. It even has > Windows code that uses VirtualAlloc to reserve actual memory. Yes, that would certainly be better that tweaking the existing code. You may want to look at ptmalloc too, it's supposed to be better at multi-threading (at least to get some ideas, the license is not 100% compatible with ours). -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Nasty Evil Memory Fragmentation fix
by Alexandre Julliard 11 Oct '01

11 Oct '01

Ove Kaaven <ovehk(a)ping.uio.no> writes: > Well, after looking at some allocator algorithms, it seems to me there's > no reason to change neither the bin sizes nor the data structures. We just > have to change the free block selection algorithm - from my reading of the > code, right now it seems to always choose the *newest* (most recently > added) free block that's big enough. By changing it to always choose the > *oldest* free block instead, we'd already have a huge improvement. > (That's essentially the effect of Gav's patch, but only for big blocks, > but I suspect that could safely be done for *all* block sizes) I don't think that's a good idea. It's pretty bad for the cache and paging system to reuse oldest blocks first. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: wine/ include/win.h include/wine/server_protoc ...
by Gerard Patel 10 Oct '01

10 Oct '01

At 06:46 PM 10/10/2001 +0200, you wrote: >No the application does a SendMessage > >096ed3f8:Call user32.SendMessageA(ffffffff,0000c00c,00000000,00000000) ret=004141aa ffffffff = Broadcast message. Take a look at this /*********************************************************************** * broadcast_message_callback * * Helper callback for broadcasting messages. */ static BOOL CALLBACK broadcast_message_callback( HWND hwnd, LPARAM lparam ) { struct send_message_info *info = (struct send_message_info *)lparam; if (!(GetWindowLongW( hwnd, GWL_STYLE ) & (WS_POPUP|WS_CAPTION))) return TRUE; GetWindowLong does not work currently between processes. There are a lot to do for the whole of Apis to work between processes yes, but there is not so much remaining for this kind of installer to work IMO. Gerard

1 0

Buisness Apps
by Nick Hudson 10 Oct '01

10 Oct '01

Greetings, Has anyone got Quick Books Pro 2001 and ACT! 2000 to run on wine? In quick books I get this error when it goes to install it: fixme:ole:CoGetClassObject CLSCTX_LOCAL_SERVER not supported! and it says that the set up failed..... in ACT! 2001 I the install stalls out when I have to say what country I am in.... I have to ctrl+alt+F1 and do a killall to get rid of it. Also its not seeing my ethernet connection.... do I need to add anything to the config file to make this work right? I was trying to install IE over the web.... As far as I can tell it doesnt see that im on the network Thanks Nick Hudson

1 0

Re: wine/ include/win.h include/wine/server_protoc ...
by Alexandre Julliard 10 Oct '01

10 Oct '01

Uwe Bonnes <bon(a)elektron.ikp.physik.tu-darmstadt.de> writes: > The WIN_FindWndPtr message flooded the xterm during install, so I supected > soething still wrong... There are still a *lot* of functions that need to be adapted for handles belonging to other processes, so yes if you run multiple processes right now you'll most likely get flooded with messages. Working on it... -- Alexandre Julliard julliard(a)winehq.com

1 0