Mail Archive Search: <+subject:/^(?:^\s*(re|sv|fwd|fw)[\[\]\d]*[:>-]+\s*)*O

Results:

References: [ +subject:/^(?:^\s*(re|sv|fwd|fw)[\[\]\d]*[:>-]+\s*)*O_DIRECT\s+address\s+restrictions\s*$/: 39 ]

Total 39 documents matching your query.

1. O_DIRECT address restrictions (score: 1): Author:; Date: Thu, 08 May 2003 15:16:22 -0700; What are the memory address restrictions when using the O_DIRECT flag with XFS? I am trying to perform I/O directly from the filesystem to a piece of mmap'ed video memory and I get the "bad address"; /archives/xfs/2003-05/msg00070.html (8,450 bytes)
2. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Fri, 9 May 2003 04:19:13 +0200; The only restriction is that the base address and the size must be an multiply of the block size of the file system (normally 4K) -Andi; /archives/xfs/2003-05/msg00072.html (9,024 bytes)
3. Re: O_DIRECT address restrictions (score: 1): Author:; Date: 08 May 2003 21:27:52 -0500; Rick, Do a man 5 on xfs and read the section on XFS_IOC_DIOINFO, if you program using this interface then should we change the alignment restrictions in xfs your programs will continue to work. There; /archives/xfs/2003-05/msg00073.html (9,800 bytes)
4. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Fri, 9 May 2003 10:55:52 -0400; Hmm interesting... when you say mmaped video memory, do you mean PCI/AGP memory *on* the video card, or system RAM? I believe some of the replies indicating that it should work were talking about sys; /archives/xfs/2003-05/msg00077.html (9,842 bytes)
5. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Fri, 09 May 2003 10:34:54 -0700; Yes, I am referring to memory physically on the video card itself (PCI), not system RAM. I am aware of the page alignment requirements of O_DIRECT under XFS, but usually I get an "invalid argument" (; /archives/xfs/2003-05/msg00082.html (11,080 bytes)
6. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Fri, 09 May 2003 21:11:30 -0700; It turns out that my problem is not specifically with XFS, but with any filesystem that supports O_DIRECT. Is it possible to perform direct I/O to addresses 0xF0000000 and above that are mmaped into; /archives/xfs/2003-05/msg00088.html (10,141 bytes)
7. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Sat, 10 May 2003 06:30:46 +0200; I don't think that will work. The direct IO path works with struct page structures in the kernel mem_map array, but these hardware mappings don't have an entry in mem_map and no struct page. In theor; /archives/xfs/2003-05/msg00089.html (9,328 bytes)
8. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Sat, 10 May 2003 20:26:46 -0700; I would be willing to upgrade memory to 4GB and/or modify the kernel to gain direct I/O functionality to/from PCI addresses 0xF0000000 through 0xF7FFFFFF. Could you recommend any resources (books, we; /archives/xfs/2003-05/msg00091.html (9,931 bytes)
9. Re: O_DIRECT address restrictions (score: 1): Author:; Date: Sun, 11 May 2003 07:07:30 +0200; The easiest way is probably to upgrade memory. Even in that case there may be still problems if someone checks PageReserved somewhere in the VM for this, but if that happens it should be quite easy t; /archives/xfs/2003-05/msg00092.html (9,062 bytes)
10. Re: O_DIRECT address restrictions (score: 1): Author: xxxx>; Date: Thu, 15 May 2003 18:32:48 -0700; It cannot be done - O_DIRECT only supports IO against legitimate RAM (memory which is described by the pageframes in mem_map[]).; /archives/xfs/2003-05/msg00164.html (8,759 bytes)
11. Re: O_DIRECT address restrictions (score: 1): Author: xxxxx>; Date: 16 May 2003 08:58:23 -0500; You mean it cannot be done with the infrastructure we have now, but it would be a cool thing to do. In fact there are legitimate applications for this. I know we have folks who move textures direct f; /archives/xfs/2003-05/msg00172.html (9,907 bytes)
12. Re: O_DIRECT address restrictions (score: 1): Author: xxxxx>; Date: Fri, 16 May 2003 16:03:09 +0200; Not very. You just have to change setup.c a bit to extent the mem_map above the physical addresses you want to copy. It cannot be done by default because such a big mem_map eats quite a lot of memor; /archives/xfs/2003-05/msg00173.html (9,029 bytes)
13. Re: O_DIRECT address restrictions (score: 1): Author: xxxxx>; Date: 19 May 2003 10:32:22 +0100; We have the abstractions to allow discontiguous mem_map, and can find out fairly easily where the PCI IO apertures are. --Stephen; /archives/xfs/2003-05/msg00190.html (9,426 bytes)
14. O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: Thu, 08 May 2003 15:16:22 -0700; What are the memory address restrictions when using the O_DIRECT flag with XFS? I am trying to perform I/O directly from the filesystem to a piece of mmap'ed video memory and I get the "bad address"; /archives/xfs/2003-05/msg00377.html (8,450 bytes)
15. Re: O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: Fri, 9 May 2003 04:19:13 +0200; The only restriction is that the base address and the size must be an multiply of the block size of the file system (normally 4K) -Andi; /archives/xfs/2003-05/msg00379.html (9,024 bytes)
16. Re: O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: 08 May 2003 21:27:52 -0500; Rick, Do a man 5 on xfs and read the section on XFS_IOC_DIOINFO, if you program using this interface then should we change the alignment restrictions in xfs your programs will continue to work. There; /archives/xfs/2003-05/msg00380.html (9,800 bytes)
17. Re: O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: Fri, 9 May 2003 10:55:52 -0400; Hmm interesting... when you say mmaped video memory, do you mean PCI/AGP memory *on* the video card, or system RAM? I believe some of the replies indicating that it should work were talking about sys; /archives/xfs/2003-05/msg00384.html (9,842 bytes)
18. Re: O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: Fri, 09 May 2003 10:34:54 -0700; Yes, I am referring to memory physically on the video card itself (PCI), not system RAM. I am aware of the page alignment requirements of O_DIRECT under XFS, but usually I get an "invalid argument" (; /archives/xfs/2003-05/msg00389.html (11,080 bytes)
19. Re: O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: Fri, 09 May 2003 21:11:30 -0700; It turns out that my problem is not specifically with XFS, but with any filesystem that supports O_DIRECT. Is it possible to perform direct I/O to addresses 0xF0000000 and above that are mmaped into; /archives/xfs/2003-05/msg00395.html (10,141 bytes)
20. Re: O_DIRECT address restrictions (score: 1): Author: xxxxxx>; Date: Sat, 10 May 2003 06:30:46 +0200; I don't think that will work. The direct IO path works with struct page structures in the kernel mem_map array, but these hardware mappings don't have an entry in mem_map and no struct page. In theor; /archives/xfs/2003-05/msg00396.html (9,328 bytes)

Current List: 1 - 20
Page: [1] [2]