RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-31 17:37:46 +00:00
Code Issues Activity Actions
59669 commits 1 branch 0 tags 230 MiB
Commit graph

59669 commits

This branch
This branch
All branches
Author SHA1 Message Date
Andreas Kling
eb78238032 Ext2FS: Don't worry about updating lookup caches in write_ext2_inode(). 
Instead, have Ext2FSInode worry about that when flushing.
Also tidy up in the inode bitmap accessors at the same time for some reason.
 2019-01-01 03:55:43 +01:00
Andreas Kling
8f9542174f Ext2FS: Don't cache a full Ext2FSInode object for non-existent inodes. 
This was a bit silly. We were always creating Ext2FSInode objects when
Ext2FSInode::get_inode() was called. They'd then sit and fatten up the
inode cache forever, despite not representing allocated inodes.

This patch consults the inode bitmap and if get_inode() is called with
an unallocated inode index, we simply cache a nullptr to represent the
fact that this index is unused.

This could be a lot better optimized, it will currently hit the disk for
every new inode index encountered.
 2019-01-01 03:37:27 +01:00
Andreas Kling
d07b08a287 FS: Don't default to having a full InodeMetadata in every Inode. 
This allows Ext2FS to keep its own ext2_inode around instead.
 2019-01-01 03:16:36 +01:00
Andreas Kling
0cb074dc73 Ext2FS: On second thought, let's not uncache the Ext2FSInodes today.. 2019-01-01 02:52:21 +01:00
Andreas Kling
741349502f Ext2FS: Free Ext2FSInodes when the last user releases them. 
The inode cache was keeping these alive forever. Added a cute little magic
trick to Retainable that calls T::one_retain_left() when the retain count
is decremented to 1.
 2019-01-01 02:38:09 +01:00
Andreas Kling
42d9f18cae Remove the cheesy block cache from DiskBackedFS. 
This should be reimplemented with proper paging support. This approach was
mostly just chewing kmalloc memory.
 2019-01-01 02:27:05 +01:00
Andreas Kling
4b6a8f8a08 Unbreak ksym loading and make reading /proc/PID/stack not crash. 2019-01-01 02:20:01 +01:00
Andreas Kling
683185e4aa MM: Allocate page tables from a separate set of physical pages. 
The old approach only worked because of an overpermissive accident.
There's now a concept of supervisor physical pages that can be allocated.
They all sit in the low 4 MB of physical memory and are identity mapped,
shared between all processes, and only ring 0 can access them.
 2019-01-01 02:09:43 +01:00
Andreas Kling
a5ffa2eec7 Add a PageDirectory::flush() that does nothing if another PD is active. 
This way callers can just flush() every time after making any modification
and the PageDirectory itself will decide if TLB invalidation is necessary.
 2018-12-31 20:52:59 +01:00
Andreas Kling
cc30407b8c Oops, errno codes need to be negative in the kernel. 2018-12-31 20:38:25 +01:00
Andreas Kling
9eca2ffd41 Make PhysicalPage eternally allocated. 2018-12-31 20:25:42 +01:00
Andreas Kling
3e37a1f5c3 Optimize PageDirectory destruction. 
Remove an extra hash lookup and only iterate over the actually-used
PhysicalPages that we need to clean up.
 2018-12-31 15:18:02 +01:00
Andreas Kling
edac1d6748 Make PageDirectory store physical pages in a HashMap. 
This container is really just there to keep a retain on the individual
PhysicalPages for each page table. A HashMap does the job with far greater
space efficiency.
 2018-12-31 15:10:12 +01:00
Andreas Kling
193ead94f8 Let PageDirectory have a PhysicalPage for the PDB instead of being the PDB. 
This is more efficient than using the wasteful kmalloc_aligned() approach
to allocation. It also allows much tighter encapsulation of the class.
 2018-12-31 14:58:03 +01:00
Andreas Kling
2239ab96e7 Tweak /bin/id output slightly. 2018-12-29 03:44:37 +01:00
Andreas Kling
eed6031064 Fix EXT2_DEBUG build. 2018-12-29 03:36:22 +01:00
Andreas Kling
0dd5279eb1 MM: Fix bug when mapping a region with a VMO with non-zero offset. 2018-12-29 03:35:48 +01:00
Andreas Kling
46ce47a984 Use the entry point address from the ELF header instead of looking up _start. 
I love these kind of dumb gotcha moments. Turns out you can find the entry
address right there in the header. :^)
 2018-12-29 03:28:55 +01:00
Andreas Kling
ab72666f48 Plug leaks in SynthFS::remove_file(). 
The process spawn stress test can now run forever. :^)
 2018-12-28 03:09:45 +01:00
Andreas Kling
3f3535213b Fix some issues uncovered by the spawn stress test. 2018-12-26 22:02:24 +01:00
Andreas Kling
f6179ad9f9 Add slightly better kmalloc_aligned() and kfree_aligned(). 
Process page directories can now actually be freed. This could definitely
be implemented in a nicer, less wasteful way, but this works for now.

The spawn stress test can now run for a lot longer but eventually dies
due to kmalloc running out of memory.
 2018-12-26 21:47:15 +01:00
Andreas Kling
55c722096d Process::create_user_process() shouldn't leak a process if exec() fails. 2018-12-26 21:04:27 +01:00
Andreas Kling
2f010e941c Unbreak the spawn stress test. 2018-12-26 20:57:51 +01:00
Andreas Kling
079889050e Merge lookup_ext2_inode() into Ext2FS::get_inode(). 2018-12-25 00:32:57 +01:00
Andreas Kling
6451b98ad4 Refactor FS::add_inode_to_directory() into Inode::add_child(). 2018-12-25 00:27:39 +01:00
Andreas Kling
4f142b86ec Refactor FS::write_inode() into Inode::write(). 2018-12-25 00:10:32 +01:00
Andreas Kling
b0db0e5de0 Get rid of Ext2FS::modify_link_count() in favor of Inode accessors. 2018-12-24 23:58:00 +01:00
Andreas Kling
673870563d Convert some FS methods to return RetainPtr<Inode>. 2018-12-24 23:45:09 +01:00
Andreas Kling
1e07ead119 Get rid of Ext2FS::is_directory_inode(). 
This was only used for assertions and will be factored out in favor of
Inode metadata checks eventually.
 2018-12-24 23:38:15 +01:00
Andreas Kling
12a6963a5d Fix bug where Vnode kept its Inode alive indefinitely. 2018-12-24 23:24:49 +01:00
Andreas Kling
033a42b580 The syncd loop can just be a lambda. 2018-12-24 23:10:48 +01:00
Andreas Kling
503e32552c Move kernel symbolication code out of init.cpp and into its own KSym files. 
Also use a simple array of { dword, const char* } for the KSyms and put the
whole shebang in kmalloc_eternal() memory. This was a fugly source of
kmalloc perma-frag.
 2018-12-24 23:01:09 +01:00
Andreas Kling
6a9fa3e41a Let sys$sigaction() fail if called with SIGKILL or SIGSTOP. 2018-12-24 22:22:35 +01:00
Andreas Kling
7bc41532be Remove FS::read_entire_inode() in favor of Inode::read_entire(). 2018-12-21 17:45:42 +01:00
Andreas Kling
04ee693925 Get rid of FS::inode_metadata() since we use Inode::metadata() everywhere. 2018-12-21 17:32:18 +01:00
Andreas Kling
8a71303827 Get rid of FS::read_inode_bytes() and use Inode::read_bytes() everywhere. 2018-12-21 17:28:16 +01:00
Andreas Kling
951ed6692b Remove InodeIdentifier::metadata(). 2018-12-21 17:16:53 +01:00
Andreas Kling
3a0a8848fb Make syscall invocations look pleasant. 
Old: Syscall::invoke(Syscall::SC_foo, (dword)arg1, (dword)arg2)
New: syscall(SC_foo, arg1, arg2)
 2018-12-21 03:02:06 +01:00
Andreas Kling
36bd53b36a Add a simple /bin/more. 2018-12-21 02:42:30 +01:00
Andreas Kling
4dd50b1f6d Get rid of three test utilities that I no longer need. 2018-12-21 02:22:21 +01:00
Andreas Kling
a9f256a745 Make the Widgets/ directory build again. 2018-12-21 02:18:16 +01:00
Andreas Kling
ec1c487dcd Yet another pass of style fixes. 2018-12-21 02:10:45 +01:00
Andreas Kling
89040cdc99 Add a "syncd" kernel process that periodically calls sync(). 2018-12-20 02:41:55 +01:00
Andreas Kling
ed7ae6c02c Add sync() syscall and a /bin/sync. 
It walks all the live Inode objects and flushes pending metadata changes
wherever needed.

This could be optimized by keeping a separate list of dirty Inodes,
but let's not get ahead of ourselves.
 2018-12-20 00:39:29 +01:00
Andreas Kling
d0f06e5f3f Automatically call Inode::flush_metadata() before an Inode is destroyed. 
Use a little template magic to have Retainable::release() call out to
T::will_be_destroyed() if such a function exists before actually calling
the destructor. This gives us full access to virtual functions in the
pre-destruction code.
 2018-12-19 22:28:09 +01:00
Andreas Kling
1f44cd9dd9 Reworked Inode to have a dirty bit and subclass-implemented flush_metadata(). 
This way we can defer disk writes as long as we like. There's no automatic
flushing happening just yet.
 2018-12-19 21:56:45 +01:00
Andreas Kling
d506c857ab Rename CoreInode to Inode. 
I don't know what I was thinking here. Clearly Inode is the right name.
 2018-12-19 21:18:28 +01:00
Andreas Kling
038d8641f9 Implement utime() along with a naive /bin/touch. 
This synchronous approach to inodes is silly, obviously. I need to rework
it so that the in-memory CoreInode object is the canonical inode, and then
we just need a sync() that flushes pending changes to disk.
 2018-12-19 21:14:55 +01:00
Andreas Kling
e03d341615 Editor: Fix build on Linux. 2018-12-09 13:48:11 +01:00
Andreas Kling
5d7ba9640c sh: Restore termios after a child process exits. 
This avoids the annoying situation that occurs when a spawned process
messes with the termios and then doesn't exit cleanly.
 2018-12-07 01:26:07 +01:00