[task]: Introduce namespaces io manage system resources #193
Conversation
Azure-stars
force-pushed
the
monolithic
branch
2 times, most recently
from
7ab1d2a to
801de3c
Compare
Azure-stars
force-pushed
the
monolithic
branch
2 times, most recently
from
572dcea to
4bd7557
Compare
Azure-stars
force-pushed
the
monolithic
branch
2 times, most recently
from
e38cb09 to
aabfe88
Compare
| RwLock::new(fd_table) | ||
| }; | ||
| def_resource! { | ||
| #[allow(non_camel_case_types)] |
There was a problem hiding this comment.
If you mean #[allow(non_camel_case_types)], when we remove it then such warnings will appear:
Structure `FD_TABLE` should have UpperCamelCase name, e.g. `FdTable`There was a problem hiding this comment.
We can add it before line 227:
arceos/modules/axns/src/lib.rs
Lines 223 to 227 in cb589bb
modules/axhal/src/mem.rs
Outdated
| @@ -90,6 +90,12 @@ fn kernel_image_regions() -> impl Iterator<Item = MemRegion> { | |||
| flags: MemRegionFlags::RESERVED | MemRegionFlags::READ, | |||
| name: ".rodata", | |||
| }, | |||
| MemRegion { | |||
| paddr: virt_to_phys((_sinit_array as usize).into()), | |||
| size: _einit_array as usize - _sinit_array as usize, | |||
There was a problem hiding this comment.
Can we use __init_array_start and __init_array_end only?
There was a problem hiding this comment.
But __init_array_end segment may be not aligned to 4K size.
.init_array : ALIGN(4K) {
_sinit_array = .;
__init_array_start = .;
*(.init_array .init_array.*)
__init_array_end = .;
. = ALIGN(4K);
_einit_array = .;
}
And the linker scripts in Unikraft are below:
OTHER SECTIONS (aligned to 0x8)
/* Constructor tables (read-only) */
. = ALIGN(0x8);
_ctors = .;
.preinit_array : {
PROVIDE_HIDDEN (__preinit_array_start = .);
KEEP (*(.preinit_array))
PROVIDE_HIDDEN (__preinit_array_end = .);
}
. = ALIGN(0x8);
.init_array : {
PROVIDE_HIDDEN (__init_array_start = .);
KEEP (*(SORT_BY_INIT_PRIORITY(.init_array.*) SORT_BY_INIT_PRIORITY(.ctors.*)))
KEEP (*(.init_array .ctors))
PROVIDE_HIDDEN (__init_array_end = .);
}
_ectors = .;
. = ALIGN(__PAGE_SIZE);
Then it enumerate all function pointers from __init_array_start to __init_array_end and invokes them.
Should we manually align this region to 4K like this?
MemRegion {
paddr: virt_to_phys((__init_array_start as usize).into()).align_down_4k(),
size: VirtAddr::from_usize(__init_array_end as usize).align_up_4k()
- VirtAddr::from_usize(__init_array_start as usize).align_down_4k(),
flags: MemRegionFlags::RESERVED | MemRegionFlags::READ,
name: ".init_array",
}There was a problem hiding this comment.
We can put the init_array section between .data and .tdata so that it will be mapped when we map _sdata.._edata.
modules/axmm/src/lib.rs
Outdated
| @@ -37,6 +37,10 @@ pub fn new_kernel_aspace() -> AxResult<AddrSpace> { | |||
| axconfig::KERNEL_ASPACE_SIZE, | |||
| )?; | |||
| for r in axhal::mem::memory_regions() { | |||
| if r.size == 0 { | |||
| info!("Skip zero-size memory region: {:?}", r); | |||
There was a problem hiding this comment.
Put the empty region check in map_linear/map_alloc/unmap/protect
There was a problem hiding this comment.
But how should we know which empty segments need to be returned normally or report error if we put empty region check in these func?
Azure-stars
force-pushed
the
monolithic
branch
from
16a9bf0 to
cb589bb
Compare
|
They will be solved in #224 |
Azure-stars
force-pushed
the
monolithic
branch
from
aabfe88 to
7e421c8
Compare
Azure-stars
force-pushed
the
monolithic
branch
3 times, most recently
from
63732c8 to
f542419
Compare
No description provided.