mirror of
https://github.com/cirosantilli/linux-kernel-module-cheat.git
synced 2026-01-28 04:24:26 +01:00
mtops.h is perfect now
This commit is contained in:
Ciro Santilli 六四事件 法轮功
204
README.adoc
204
README.adoc
@@ -1005,12 +1005,6 @@ cd linux-kernel-module-cheat
|
||||
./build --download-dependencies userland-host
|
||||
....
|
||||
|
||||
If you modify a dependency that is not currently considered such as a header file, force the rebuild with:
|
||||
|
||||
....
|
||||
./build --force-rebuild
|
||||
....
|
||||
|
||||
The `build` scripts inside link:userland/[] are just symlinks to link:build-userland-in-tree[] which you can also use from toplevel as:
|
||||
|
||||
....
|
||||
@@ -1021,6 +1015,22 @@ The `build` scripts inside link:userland/[] are just symlinks to link:build-user
|
||||
|
||||
which is in turn just a thin wrapper around link:build-userland[], so you can use any option supported by that script freely.
|
||||
|
||||
Pass custom compiler options with:
|
||||
|
||||
....
|
||||
./build-userland --ccflags='-foptimize-sibling-calls -foptimize-strlen' --force-rebuild
|
||||
....
|
||||
|
||||
Here we used `--force-rebuild` to force rebuild since the sources weren't modified since the last build.
|
||||
|
||||
Some CLI options have more specialized flags, e.g. `-O` optimization level:
|
||||
|
||||
....
|
||||
./build-userland --optimization-level 3 --force-rebuild
|
||||
....
|
||||
|
||||
See also <<user-mode-static-executables>> for `--static`.
|
||||
|
||||
Do a more clean out-of-tree build and run the program instead:
|
||||
|
||||
....
|
||||
@@ -3407,13 +3417,14 @@ Let's run link:userland/c/print_argv.c[] built with the Buildroot toolchain on Q
|
||||
./build user-mode-qemu
|
||||
./run \
|
||||
--userland c/print_argv \
|
||||
--userland-args 'asdf "qw er"' \
|
||||
--userland-args='asdf "qw er"' \
|
||||
;
|
||||
....
|
||||
|
||||
Output:
|
||||
|
||||
....
|
||||
/path/to/linux-kernel-module-cheat/out/userland/default/x86_64/c/print_argv.out
|
||||
asdf
|
||||
qw er
|
||||
....
|
||||
@@ -3428,33 +3439,72 @@ If you modify the userland programs, rebuild simply with:
|
||||
./build-userland
|
||||
....
|
||||
|
||||
==== User mode GDB step debug
|
||||
==== User mode GDB
|
||||
|
||||
The commands are analogous to full system <<gdb>>, e.g. without tmux:
|
||||
It's nice when <<gdb,the obvious>> just works, right?
|
||||
|
||||
....
|
||||
./run \
|
||||
--arch aarch64 \
|
||||
--userland c/print_argv \
|
||||
--userland-args 'asdf "qw er"' \
|
||||
--wait-gdb \
|
||||
;
|
||||
....
|
||||
|
||||
and on another shell:
|
||||
|
||||
....
|
||||
./run-gdb \
|
||||
--arch aarch64 \
|
||||
--userland c/print_argv \
|
||||
main
|
||||
main \
|
||||
;
|
||||
....
|
||||
|
||||
or with <<tmux>>:
|
||||
Or alternatively, if you are using <<tmux>>, do everything in one go with:
|
||||
|
||||
....
|
||||
./run \
|
||||
--arch aarch64 \
|
||||
--tmux-args main \
|
||||
--userland c/print_argv \
|
||||
--userland-args 'asdf "qw er"' \
|
||||
--tmux-args main \
|
||||
--wait-gdb \
|
||||
;
|
||||
....
|
||||
|
||||
To stop at the very first instruction of a freestanding program, just use `--no-continue` TODO example.
|
||||
|
||||
=== User mode tests
|
||||
|
||||
Automatically run userland tests that can be run in user mode simulation, and check that they exit with status 0:
|
||||
|
||||
....
|
||||
./build --all-archs test-user-mode
|
||||
./test-user-mode --all-archs --all-emulators
|
||||
....
|
||||
|
||||
Or just for QEMU:
|
||||
|
||||
....
|
||||
./build --all-archs test-user-mode-qemu
|
||||
./test-user-mode --all-archs --emulator qemu
|
||||
....
|
||||
|
||||
Source: link:test-user-mode[]
|
||||
|
||||
This script skips a manually configured list of tests, notably:
|
||||
|
||||
* tests that depend on a full running kernel and cannot be run in user mode simulation, e.g. those that rely on kernel modules
|
||||
* tests that require user interaction
|
||||
* tests that take perceptible ammounts of time
|
||||
* known bugs we didn't have time to fix ;-)
|
||||
|
||||
The gem5 tests require building statically with build id `static`, see also: <<gem5-syscall-emulation-mode>>. TODO automate this better.
|
||||
|
||||
See: <<test-this-repo>> for more useful testing tips.
|
||||
|
||||
==== User mode with host toolchain and QEMU
|
||||
|
||||
If you are lazy to built the Buildroot toolchain and QEMU, you can get away on Ubuntu 18.04 with just:
|
||||
@@ -3484,13 +3534,15 @@ We must pass this to `./run` as well because QEMU must know which dynamic librar
|
||||
|
||||
This present the usual trade-offs of using prebuilts as mentioned at: <<prebuilt>>.
|
||||
|
||||
When you build with the native host toolchain, you can also execute many of the executables directly natively on the host: <<userland-setup-getting-started-natively>>.
|
||||
|
||||
==== User mode simulation with glibc
|
||||
|
||||
At 125d14805f769104f93c510bedaa685a52ec025d we <<libc-choice,moved Buildroot from uClibc to glibc>>, and caused some user mode pain, which we document here.
|
||||
|
||||
===== FATAL: kernel too old
|
||||
|
||||
Happens on all gem5 setups, but not on QEMU on Ubuntu 18.04 host.
|
||||
Happens on all gem5 <<user-mode-simulation>> setups, but not on QEMU on Ubuntu 18.04 host.
|
||||
|
||||
glibc has a check for kernel version, likely obtained from the `uname` syscall, and if the kernel is not new enough, it quits.
|
||||
|
||||
@@ -3504,11 +3556,12 @@ We don't have this failure for QEMU, only gem5. QEMU by default copies the host
|
||||
|
||||
Source: link:userland/posix/uname.c[].
|
||||
|
||||
gem5 does not have such runtime configuration, but the error can be worked around for now by patching the hardcoded Linux version as mentioned at: https://stackoverflow.com/questions/48959349/how-to-solve-fatal-kernel-too-old-when-running-gem5-in-syscall-emulation-se-m to be a recent Linux version such as `v4.17.0`.
|
||||
|
||||
The QEMU source that does this is at: https://github.com/qemu/qemu/blob/v3.1.0/linux-user/syscall.c#L8931
|
||||
|
||||
In gem5, there are tons of missing syscalls, and that number currently just gets bumped up randomly from time to time when someone gets fed up:
|
||||
|
||||
* https://stackoverflow.com/questions/48959349/how-to-solve-fatal-kernel-too-old-when-running-gem5-in-syscall-emulation-se-m
|
||||
* https://stackoverflow.com/questions/53085048/how-to-compile-and-run-an-executable-in-gem5-syscall-emulation-mode-with-se-py/53085049#53085049
|
||||
* https://gem5-review.googlesource.com/c/public/gem5/+/15855
|
||||
|
||||
@@ -3597,43 +3650,6 @@ it fails with:
|
||||
ld: cannot find -lopenblas
|
||||
....
|
||||
|
||||
==== User mode GDB
|
||||
|
||||
It's nice when <<gdb,the obvious>> just works, right?
|
||||
|
||||
....
|
||||
./run \
|
||||
--arch aarch64 \
|
||||
--userland c/print_argv \
|
||||
--userland-args 'asdf "qw er"' \
|
||||
--wait-gdb \
|
||||
;
|
||||
....
|
||||
|
||||
and on another shell:
|
||||
|
||||
....
|
||||
./run-gdb \
|
||||
--arch aarch64 \
|
||||
--userland c/print_argv \
|
||||
main \
|
||||
;
|
||||
....
|
||||
|
||||
Or alternatively, if you are using <<tmux>>, do everything in one go with:
|
||||
|
||||
....
|
||||
./run \
|
||||
--arch aarch64 \
|
||||
--tmux-args main \
|
||||
--userland c/print_argv \
|
||||
--userland-args 'asdf "qw er"' \
|
||||
--wait-gdb \
|
||||
;
|
||||
....
|
||||
|
||||
To stop at the very first instruction of a freestanding program, just use `--no-continue` TODO example.
|
||||
|
||||
=== gem5 syscall emulation mode
|
||||
|
||||
Less robust than QEMU's, but still usable:
|
||||
@@ -3671,7 +3687,7 @@ So let's just play with some static ones:
|
||||
|
||||
TODO: how to escape spaces on the command line arguments?
|
||||
|
||||
<<user-mode-gdb-step-debug,GDB step debug>> also works normally on gem5:
|
||||
<<user-mode-gdb,GDB step debug>> also works normally on gem5:
|
||||
|
||||
....
|
||||
./run \
|
||||
@@ -3775,30 +3791,6 @@ Result on <<p51>> at bad30f513c46c1b0995d3a10c0d9bc2a33dc4fa0:
|
||||
* QEMU user: 45 seconds
|
||||
* QEMU full system: 223 seconds
|
||||
|
||||
=== User mode tests
|
||||
|
||||
Automatically run non-interactive userland tests that can be run in user mode simulation:
|
||||
|
||||
....
|
||||
./build --all-archs test-user-mode
|
||||
./test-user-mode --all-archs --all-emulators
|
||||
....
|
||||
|
||||
Or just for QEMU:
|
||||
|
||||
....
|
||||
./build --all-archs test-user-mode-qemu
|
||||
./test-user-mode --all-archs --emulator qemu
|
||||
....
|
||||
|
||||
Source: link:test-user-mode[]
|
||||
|
||||
This testing excludes notably kernel module tests which depend on a full running kernel.
|
||||
|
||||
The gem5 tests require building statically with build id `static`, see also: <<gem5-syscall-emulation-mode>>. TODO automate this better.
|
||||
|
||||
See: <<test-this-repo>> for more useful testing tips.
|
||||
|
||||
== Kernel module utilities
|
||||
|
||||
=== insmod
|
||||
@@ -9908,21 +9900,30 @@ https://stackoverflow.com/questions/6147242/heap-vs-binary-search-tree-bst/29548
|
||||
Usage:
|
||||
|
||||
....
|
||||
./build-userland \
|
||||
--arch aarch64 \
|
||||
--ccflagg='-DLKMC_M5OPS_ENABLE=1' \
|
||||
--force-build cpp/bst_vs_heap \
|
||||
--static \
|
||||
;
|
||||
./run \
|
||||
--arch aarch64 \
|
||||
--eval-after './gem5.sh' \
|
||||
--emulator gem5 \
|
||||
--gem5-readfile './bst_vs_heap.out' \
|
||||
--static \
|
||||
--userland cpp/bst_vs_heap \
|
||||
;
|
||||
./bst-vs-heap --arch aarch64 --emulator gem5 > bst_vs_heap.dat
|
||||
./bst-vs-heap --arch aarch64 > bst_vs_heap.dat
|
||||
./bst-vs-heap.gnuplot
|
||||
xdg-open bst-vs-heap.tmp.png
|
||||
....
|
||||
|
||||
and then feed `bst_vs_heap.dat` into: https://github.com/cirosantilli/cpp-cheat/blob/9d0f77792fc8e55b20b6ee32018761ef3c5a3f2f/cpp/interactive/bst_vs_heap.gnuplot
|
||||
|
||||
Sources:
|
||||
|
||||
* link:userland/cpp/bst_vs_heap.cpp[]
|
||||
* link:bst-vs-heap[]
|
||||
* link:userland/bst_vs_heap.cpp[]
|
||||
* link:bst-vs-heap.gnuplot[]
|
||||
|
||||
Tested on e70103b9b32e6e33dbab9eaf2ff00c358f55d8db + 1 with the workaround patch mentioned at: <<fatal-kernel-too-old>>.
|
||||
|
||||
===== BLAS
|
||||
|
||||
@@ -10585,29 +10586,44 @@ adsf
|
||||
|
||||
==== m5ops instructions
|
||||
|
||||
The executable `./m5ops.out` illustrates how to hard code with inline assembly the m5ops that you are most likely to hack into the benchmark you are analysing:
|
||||
gem5 allocates some magic instructions on unused instruction encodings for convenient guest instrumentation.
|
||||
|
||||
Those instructions are exposed through the <<m5>> in tree executable.
|
||||
|
||||
To make things simpler to understand, you can play around with our own minimized educational `m5` subset link:userland/c/m5ops.c[].
|
||||
|
||||
The instructions used by `./c/m5ops.out` are present in link:lkmc/m5ops.h[] in a very simple to understand and reuse inline assembly form.
|
||||
|
||||
To use that file, first rebuild `m5ops.out` with the m5ops instructions enabled and install it on the root filesystem:
|
||||
|
||||
....
|
||||
./build-userland \
|
||||
--arch aarch64 \
|
||||
--ccflags='-DLKMC_M5OPS_ENABLE=1' \
|
||||
--force-build c/m5ops \
|
||||
--static \
|
||||
;
|
||||
./build-buildroot --arch aarch64
|
||||
....
|
||||
|
||||
We don't enable `-DLKMC_M5OPS_ENABLE=1` by default on userland executables because we try to use a single image for both gem5, QEMU and <<userland-setup-getting-started-natively,native>>, and those instructions would break the latter two. We enable it in the <<baremetal-setup>> by default since we already have different images for QEMU and gem5 there.
|
||||
|
||||
Then, from inside <<gem5-buildroot-setup>>, test it out with:
|
||||
|
||||
....
|
||||
# checkpoint
|
||||
./m5ops.out c
|
||||
./c/m5ops.out c
|
||||
|
||||
# dumpstats
|
||||
./m5ops.out d
|
||||
./c/m5ops.out d
|
||||
|
||||
# exit
|
||||
./m5ops.out e
|
||||
./c/m5ops.out e
|
||||
|
||||
# dump resetstats
|
||||
./m5ops.out r
|
||||
./c/m5ops.out r
|
||||
....
|
||||
|
||||
Sources:
|
||||
|
||||
* link:userland/m5ops.h[]
|
||||
* link:userland/m5ops.c[]
|
||||
|
||||
That executable is of course a subset of <<m5>> and useless by itself: its goal is only illustrate how to hardcode some <<m5ops>> yourself as one-liners.
|
||||
|
||||
In theory, the cleanest way to add m5ops to your benchmarks would be to do exactly what the `m5` tool does:
|
||||
|
||||
* include link:https://github.com/gem5/gem5/blob/05c4c2b566ce351ab217b2bd7035562aa7a76570/include/gem5/asm/generic/m5ops.h[`include/gem5/asm/generic/m5ops.h`]
|
||||
|
||||
Reference in New Issue
Block a user