sambuc/codezero
1
0
Fork 0
mirror of https://github.com/drasko/codezero.git synced 2026-01-12 10:53:16 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity
152 Commits 1 Branch 0 Tags
dada3e0b2cdf18426b23b9007bfa202b55873a19
Commit Graph

152 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Bahadir Balban
dada3e0b2c Added copy_vma() function. 
Sorting out return_from_fork() and kernel stack setup for the child.
 2008-08-21 12:18:38 +03:00
Bahadir Balban
69db3a04c0 Towards implementing fork. 
Issue is that shadow object references from original objects are into
the links rather than the objects.
 2008-08-19 18:03:23 +03:00
Bahadir Balban
f436b44e81 Changes in the README, changes to fork template. 2008-08-16 13:01:18 +03:00
Bahadir Balban
d434ad4b40 Changes towards adding clone/fork 
new file:   src/clone.c
 2008-08-05 16:56:55 +03:00
Bahadir Balban
d1170211e3 Commented out printfs for pager/vfs interaction 2008-05-30 11:18:16 +03:00
Bahadir Balban
f561d885d5 Flushing pages to vfs will work, but the issue is that while vfs is 
serving mm0, if it page faults, system deadlocks because mm0 is waiting to be served by vfs.

FIX: To fix this, mm0 will need to fork itself and keep a separate thread solely for
page fault handling.
 2008-05-28 23:37:41 +03:00
Bahadir Balban
98888a94b5 Modified scripts 2008-05-27 10:17:46 +03:00
Bahadir Balban
fd787479c3 Fixed minor faults during close path. 
FIXME:
write/close/open/read sequence does not read the initially written data.
Investigate.
 2008-05-23 05:42:26 +00:00
Bahadir Balban
6be310f460 Added sys_close and sys_fsync 
Untested.
 2008-05-23 03:59:18 +00:00
Bahadir Balban
00b47117b0 Added sys_stat, sys_fstat and their libposix glue. 2008-05-22 11:21:27 +00:00
Bahadir Balban
bd8b182f1b Added syscall to update filestats (size) from pager. 
After flushing written/truncated pages, stats are updated
so that vfs is synced with new file info such as its size.
 2008-05-21 10:23:11 +00:00
Bahadir Balban
5efece98cd More changes 2008-05-03 15:04:16 +01:00
Bahadir Balban
0557fa8910 Gotta find a way to pass on updated file size before write_file_pages. 2008-04-26 01:03:11 +01:00
Bahadir Balban
f183b3c15e Still towards close/flush. 2008-04-25 23:06:23 +01:00
Bahadir Balban
c9b9f9d111 Towards sys_close()/sys_flush() 2008-04-25 22:14:59 +01:00
Bahadir Balban
488d9e6271 Fixed the issue that page copy offsets were not correct. 
When copying file pages back and forth, we need both dst and src
offsets since the cursor offset and read/write buffer page offsets
are different. This is now fixed.

Also done some mods to close().
 2008-04-25 17:07:49 +01:00
Bahadir Balban
ba725168a0 Fixed lseek. Updates to sys_read/sys_write 
read and write are both called by test0. But they don't yet
work as expected. An error is in one or both. To be investigated.
 2008-04-24 00:09:49 +01:00
Bahadir Balban
0629ba3c1a sys_write() "seems to" work. Also added support for sys_lseek(). 2008-04-23 14:14:45 +01:00
Bahadir Balban
59bdaf22ec Towards user buffer checking 2008-04-22 14:18:12 +01:00
Bahadir Balban
1270cfe3b3 open file descriptor references to vm files weren't properly initialised. 
This fixes the issue. sys_read/write yet to be inspected.
 2008-04-22 01:41:18 +01:00
Bahadir Balban
e8bb529dcb Now we save/restore utcb registers if we do a second ipc before returning the first. 2008-04-22 00:30:07 +01:00
Bahadir Balban
80470e2581 Adding reply phase to pager_sys_open() 
So that open() errors on pager's side can be noticed by vfs.
 2008-04-21 15:30:40 +01:00
Bahadir Balban
a674cf2b27 Recompiled with write_cache_pages(). Implementation to be re-evaluated. 2008-04-21 00:31:16 +01:00
Bahadir Balban
e3cc14515c mm0 compiles, still previous issues on. 2008-04-20 23:34:25 +01:00
Bahadir Balban
d8d65a6301 Efforts towards sys_write. 
- Added logic to calculate which pages are in the file and which are new
  for writing.
- Nearly finished the template for sys_write.
- Need to implement write_cache_pages().
- Need to distinguish dirty/clean pages and implement flushing dirty
  pages back to vfs.
 2008-04-20 17:33:05 +01:00
Bahadir Balban
9992d100d7 Few fixes in libposix error checking. 
size_t is a non-negative type and negative error checking didn't work. Now fixed.
Fixed various issues with reading file pages in the sys_read() path.
 2008-04-20 02:12:53 +01:00
Bahadir Balban
8b3fedc18d Towards working sys_read/sys_write. 2008-04-19 15:42:13 +01:00
Bahadir Balban
9f4e400fed Minor updates to README 2008-04-19 12:28:49 +01:00
Bahadir Balban
df0eccf7b3 A file can be created. 
Now up to sys_write() in pager, which needs to be filled in.
 2008-04-19 01:27:41 +01:00
Bahadir Balban
ce26835968 fd wasn't returned in reorganised sys_open. Fixed. 2008-04-19 00:55:11 +01:00
Bahadir Balban
8c4c436925 Reorganised sys_open. 
vfs_create and mknod now returns the newly created vnode.
(which might be used by upper layers).
 2008-04-19 00:40:48 +01:00
Bahadir Balban
f7163b7e93 Added file io test to test0. Added close call to libposix. 2008-04-18 21:17:09 +01:00
Bahadir Balban
cff7a505e8 Changed time representation to posix style struct timeval/ gettimeofday. 2008-04-18 13:58:37 +01:00
Bahadir Balban
5563cc1c6b Minor changes on sys_time 
A get/set flag determines whether to read or write time.
 2008-04-18 01:03:39 +01:00
Bahadir Balban
a87914910c Added a new system call sys_timer. 
sys_timer accumulates timer ticks into seconds, minutes, hours and days.
It's left to the user to calculate from days into a date. It is not yet
known if the calculation is even roughly correct.

Reduced 2 kmem_reclaim/grant calls into one kmem_control call.
 2008-04-18 00:46:29 +01:00
Bahadir Balban
73058dc249 Added sys_close() but its not done yet. 
close() needs to flush dirty buffers.
pager needs to have read/write support properly implemented.
open() needs to record mode and access times.
The need for access times means we need rtc and time implementation.
Also need to add stat() access() etc.
 2008-04-17 21:47:37 +01:00
Bahadir Balban
58033e7927 CHDIR works. 
Lookups on different current directory, including /./././/// works as expected.
 2008-04-16 16:00:17 +01:00
Bahadir Balban
730e7c210f Added resolving of current directory in case lookup starts from current dir. 2008-04-16 14:48:05 +01:00
Bahadir Balban
9d32f840c0 Open wasn't returning the fd but 0 on success :-S. fixed. 2008-04-16 00:34:11 +01:00
Bahadir Balban
e722ee0115 Root was a child of itself, removed that relationship. 
Root is only a parent of itself, it shouldn't show up in its children.
 2008-04-16 00:20:59 +01:00
Bahadir Balban
ab588c279b Fixed the component-consumption-every-lookup problem. 
Now components only consumed on child lookup recursions.
 2008-04-16 00:04:54 +01:00
Bahadir Balban
8a3da9c709 README update 
modified:   README
 2008-04-15 23:57:21 +01:00
Bahadir Balban
9a66893288 lookup return value was changed from -ENOENT to 0 when vnode not found, 
Fixed a point in code that expected 0.

Next: lookups should not consume from path every time called but only
when moving to children to lookup.
 2008-04-15 23:45:46 +01:00
Bahadir Balban
577a2bffe4 Changed name of cml configuration 2008-04-15 19:43:35 +01:00
Bahadir Balban
9ba6638d01 Added new path parsing functions that work cleaner, better. 
Input paths are now parsed at the beginning and components
put into an ordered linked list headed by struct pathdata.
Lookup functions use these components to look up vnodes.
 2008-04-15 19:36:07 +01:00
Bahadir Balban
fb249b0c5b Fixed wrong directory name issue, strlen() was off by one. 2008-04-15 01:51:13 +01:00
Bahadir Balban
a9420d3dc9 mkdir almost working. 
Added changes to pathname lookup code so that the root
directory special case is handled properly.
 2008-04-15 00:51:58 +01:00
Bahadir Balban
287b7705da Towards working mkdir. 2008-04-14 14:38:04 +01:00
Bahadir Balban
d6d97876bb Page fault handling fix. 
Factored out mapping of the physical page as the final generic code
after all fault-specific handling is done.

Fixed the error that zero page didn't have an owner (devzero).

Fixed the error that struct dirent did not have the record length
field as u16 as expected by userspace.
 2008-04-14 00:09:57 +01:00
Bahadir Balban
7f815dd2dd Fixed handling of pages that are SHARED and ANONYMOUS 
Normally anon && shared pages need to COW once and any newer
accesses by other sharing processes should simply map the COW'ed
pages to themselves as writeable. We didn't have that but instead
every ANON && SHARED write fault was going into the COW path, causing
the same offset'ed page to be COW'ed twice.

Next: Factor out the l4_map code to a common point at the end of all
fault handling paths.
 2008-04-13 17:03:11 +01:00