There is a problem with the current elf loader for risc-v: when a pair of
PCREL_HI20 / LO12 relocations are encountered, it is assumed that these
will follow each other immediately, as follows:
label:
auipc a0, %pcrel_hi(symbol) // R_RISCV_PCREL_HI20
load/store a0, %pcrel_lo(label)(a0) // R_RISCV_PCREL_LO12_I/S
With this assumption, the hi/lo relocations are both done when a hi20
relocation entry is encountered, first to the current instruction (addr)
and to the next instruction (addr + 4).
However, this assumption is wrong. There is nothing in the elf relocation
specification[1] that mandates this. Thus, the hi/lo relocation always
needs to first fixup the hi-part, and when the lo-part is encountered, it
needs to find the corresponding hi relocation entry, via the given "label".
This necessitates (re-)visiting the relocation entries for the current
section as well as looking for "label" in the symbol table.
The NuttX elf loader does not allow such operations to be done in the
machine specific part, so this patch fixes the relocation issue by
introducing an architecture specific cache for the hi20 relocation and
symbol table entries. When a lo12 relocation is encountered, the cache
can be consulted to find the hi20 part.
[1] https://github.com/riscv-non-isa/riscv-elf-psabi-doc/blob/master/riscv-elf.adoc
This option enables architecture-specific memory copy for dynamic code loading.
For example, Ambiq has MRAM regions for instruction which can't load by
the memcpy directly.
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
Load all sections to LMA not VMA, so the startup code(e.g. start.S) need
relocate .data section to the final address(VMA) and zero .bss section by self.
For example, SiFli and Actions: Background: Device with small sram,
Bootloader run in sram and psram, need boot to Application, with memory overlap
and without XIP. VMA of .data is in "psram" and LMA in "rom", if not enable
`ELF_LOADTO_LMA`, ELF loader will load the section to VMA (will fill bootloader
itself).
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
From posix spec, if set-user-ID bit is set in the file permissions,
then the effective user ID of the new process shall be set to the
user ID of the new process image file.
Let's ignore whether ST_NOSUID is set on the mounted file system.
https://pubs.opengroup.org/onlinepubs/007904875/functions/exec.html
test step:
hello example build as a module and call geteuid and getegid API.
then set file binary set-user-ID bit on the host.
$ chmod +s apps/bin/hello
nsh> mount -t hostfs -o fs=. /data
nsh> ls -l /data/apps/bin/hello
-rwsrwsr-x 1000 1000 9264 /data/apps/bin/hello
nsh> /data/apps/bin/hello
geteuid:1000
getegid:1000
Signed-off-by: fangxinyong <fangxinyong@xiaomi.com>
1. Update all CMakeLists.txt to adapt to new layout
2. Fix cmake build break
3. Update all new file license
4. Fully compatible with current compilation environment(use configure.sh or cmake as you choose)
------------------
How to test
From within nuttx/. Configure:
cmake -B build -DBOARD_CONFIG=sim/nsh -GNinja
cmake -B build -DBOARD_CONFIG=sim:nsh -GNinja
cmake -B build -DBOARD_CONFIG=sabre-6quad/smp -GNinja
cmake -B build -DBOARD_CONFIG=lm3s6965-ek/qemu-flat -GNinja
(or full path in custom board) :
cmake -B build -DBOARD_CONFIG=$PWD/boards/sim/sim/sim/configs/nsh -GNinja
This uses ninja generator (install with sudo apt install ninja-build). To build:
$ cmake --build build
menuconfig:
$ cmake --build build -t menuconfig
--------------------------
2. cmake/build: reformat the cmake style by cmake-format
https://github.com/cheshirekow/cmake_format
$ pip install cmakelang
$ for i in `find -name CMakeLists.txt`;do cmake-format $i -o $i;done
$ for i in `find -name *\.cmake`;do cmake-format $i -o $i;done
Co-authored-by: Matias N <matias@protobits.dev>
Signed-off-by: chao an <anchao@xiaomi.com>
The following changes make it possible for Nuttx to load binaries in ELF format which are fully linked.
The change does not include the necessary modifications to produce such binaries. In order to build an applicable binary:
- The userspace applications linker script (`gnu-elf.ld`) needs to modified so the data and text section origin's match those setup by the address environment.
- The makefile used, in `apps/import/Make.defs` needs to remove the `-r` LDELFFLAG.
Store the old environment in a local context so another temporary address
environment can be selected. This can happen especially when a process
is being loaded (the new process's mappings are temporarily instantiated)
and and interrupt occurs.
Instead of using a volatile storage for the address environment in the
binfmt / loadinfo structures, always allocate the address environment
from kheap.
This serves two purposes:
- If the task creation fails, any kernel thread that depends on the
address environment created during task creation will not lose their
mappings (because they hold a reference to it)
- The current address environment variable (g_addrenv) will NEVER contain
a stale / incorrect value
- Releasing the address environment is simplified as any pointer given
to addrenv_drop() can be assumed to be heap memory
- Makes the kludge function addrenv_clear_current irrelevant, as the
system will NEVER have invalid mappings any more
The function is not relevant any longer, remove it. Also remove
save_addrenv_t, the parameter taken by up_addrenv_restore.
Implement addrenv_select() / addrenv_restore() to handle the temporary
instantiation of address environments, e.g. when a process is being
created.
Same as with group_free(), there is no need to instantiate the address
environment to destroy it.
The only problem was the ARM implementation modified the L1 mappings
in up_addrenv_destroy(), which it no longer does.
In the past, header file paths were generated by the incdir command
Now they are generated by concatenating environment variables
In this way, when executing makefile, no shell command will be executed,
it will improve the speed of executing makfile
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
Fixes minor issue with a trace (build failure with -Werror):
libelf/libelf_symbols.c:310:41: error: pointer of type 'void *' used in arithmetic [-Werror=pointer-arith]
310 | (uintptr_t)(sym->st_value + symbol->sym_value));
Implement a generic access rights modification procedure instead
of the procedures that only do one thing (enable/disable write)
to one section (text).
When the .elf file is loaded from disk, the kernel must be given write
access to the allocated .text section in the task's address environment.
The access is removed after the elf is loaded and relocations are done.
NOTE:
The reason this works for the ARM implementation, is that the ARM MMU
can be configured to give write access for the privileged mode, but
revoke write access for the user mode.
Regardless, it would be smart to revoke write access even for the
kernel, when the kernel does not need it. This framework allows doing
that, if someone wishes to take up the task.
Basically, mirror the following two commits from modlib.
It's shame we have two copies of elf loaders.
```
commit 51490bad55
Author: YAMAMOTO Takashi <yamamoto@midokura.com>
Date: Wed Apr 14 17:07:39 2021 +0900
modlib: Implement sh_addralign handling
I've seen a module with 16 bytes .rodata alignment for xmm operations.
It was getting SEGV on sim/Linux because of the alignment issue.
The same module binary seems working fine after applying this patch.
Also, tested on sim/macOS and esp32 on qemu,
using a module with an artificially large alignment. (64 bytes)
```
```
commit 418e11b8b3
Author: YAMAMOTO Takashi <yamamoto@midokura.com>
Date: Thu Apr 15 11:33:48 2021 +0900
modlib: Always use separate allocation for text and data
Pros:
* Reduce code differences
* Smaller allocations for !CONFIG_ARCH_USE_MODULE_TEXT
Cons:
* Likely to use more memory for !CONFIG_ARCH_USE_MODULE_TEXT in total
Tested with:
* sim:module on macOS
* esp32-devkit:nsh + CONFIG_MODULE on qemu
* lm3s6965-ek:qemu-protected + CONFIG_EXAMPLES_SOTEST on qemu
```
Ville Juven