3rd-party/linux-kernel-module-cheat
1
0
Fork 0
mirror of https://github.com/cirosantilli/linux-kernel-module-cheat.git synced 2026-01-25 19:21:35 +01:00
Code Issues Releases Wiki Activity

304bf3dbe9

Browse Source
This commit is contained in:
Ciro Santilli 六四事件 法轮功
2020-10-30 00:00:00 +00:00
parent 37627d90aa
commit 79c2baf7da
1 changed files with 32 additions and 268 deletions
Download Patch File Download Diff File Expand all files Collapse all files

300
index.html
View File

@@ -5,7 +5,7 @@
<meta http-equiv="X-UA-Compatible" content="IE=edge"> <meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1.0"> <meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta name="generator" content="Asciidoctor 2.0.10"> <meta name="generator" content="Asciidoctor 2.0.10">
<meta name="description" content="The perfect emulation setup to study and develop the &lt;&lt;linux-kernel&gt;&gt; v5.4.3, kernel modules, &lt;&lt;qemu-buildroot-setup,QEMU&gt;&gt;, &lt;&lt;gem5-buildroot-setup,gem5&gt;&gt; and x86_64, ARMv7 and ARMv8 &lt;&lt;userland-assembly,userland&gt;&gt; and &lt;&lt;baremetal-setup,baremetal&gt;&gt; assembly, &lt;&lt;c,ANSI C&gt;&gt;, &lt;&lt;cpp,C++&gt;&gt; and &lt;&lt;posix,POSIX&gt;&gt;. &lt;&lt;gdb&gt;&gt; and &lt;&lt;kgdb&gt;&gt; just work. Powered by &lt;&lt;about-the-qemu-buildroot-setup,Buildroot&gt;&gt; and &lt;&lt;about-the-baremetal-setup,crosstool-NG&gt;&gt;. Highly automated. Thoroughly documented. Automated &lt;&lt;test-this-repo,tests&gt;&gt;. "Tested" in an Ubuntu 20.04 host."> <meta name="description" content="The perfect emulation setup to study and develop the &lt;&lt;linux-kernel&gt;&gt; v5.9.2, kernel modules, &lt;&lt;qemu-buildroot-setup,QEMU&gt;&gt;, &lt;&lt;gem5-buildroot-setup,gem5&gt;&gt; and x86_64, ARMv7 and ARMv8 &lt;&lt;userland-assembly,userland&gt;&gt; and &lt;&lt;baremetal-setup,baremetal&gt;&gt; assembly, &lt;&lt;c,ANSI C&gt;&gt;, &lt;&lt;cpp,C++&gt;&gt; and &lt;&lt;posix,POSIX&gt;&gt;. &lt;&lt;gdb&gt;&gt; and &lt;&lt;kgdb&gt;&gt; just work. Powered by &lt;&lt;about-the-qemu-buildroot-setup,Buildroot&gt;&gt; and &lt;&lt;about-the-baremetal-setup,crosstool-NG&gt;&gt;. Highly automated. Thoroughly documented. Automated &lt;&lt;test-this-repo,tests&gt;&gt;. "Tested" in an Ubuntu 20.04 host.">
<title>Linux Kernel Module Cheat</title> <title>Linux Kernel Module Cheat</title>
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Open+Sans:300,300italic,400,400italic,600,600italic%7CNoto+Serif:400,400italic,700,700italic%7CDroid+Sans+Mono:400,700"> <link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Open+Sans:300,300italic,400,400italic,600,600italic%7CNoto+Serif:400,400italic,700,700italic%7CDroid+Sans+Mono:400,700">
<style> <style>
@@ -456,7 +456,7 @@ pre{ white-space:pre }
<p><a href="https://zenodo.org/badge/latestdoi/64534859"><span class="image"><img src="https://zenodo.org/badge/64534859.svg" alt="64534859"></span></a></p> <p><a href="https://zenodo.org/badge/latestdoi/64534859"><span class="image"><img src="https://zenodo.org/badge/64534859.svg" alt="64534859"></span></a></p>
</div> </div>
<div class="paragraph"> <div class="paragraph">
<p>The perfect emulation setup to study and develop the <a href="#linux-kernel">Linux kernel</a> v5.4.3, kernel modules, <a href="#qemu-buildroot-setup">QEMU</a>, <a href="#gem5-buildroot-setup">gem5</a> and x86_64, ARMv7 and ARMv8 <a href="#userland-assembly">userland</a> and <a href="#baremetal-setup">baremetal</a> assembly, <a href="#c">ANSI C</a>, <a href="#cpp">C++</a> and <a href="#posix">POSIX</a>. <a href="#gdb">GDB step debug</a> and <a href="#kgdb">KGDB</a> just work. Powered by <a href="#about-the-qemu-buildroot-setup">Buildroot</a> and <a href="#about-the-baremetal-setup">crosstool-NG</a>. Highly automated. Thoroughly documented. Automated <a href="#test-this-repo">tests</a>. "Tested" in an Ubuntu 20.04 host.</p> <p>The perfect emulation setup to study and develop the <a href="#linux-kernel">Linux kernel</a> v5.9.2, kernel modules, <a href="#qemu-buildroot-setup">QEMU</a>, <a href="#gem5-buildroot-setup">gem5</a> and x86_64, ARMv7 and ARMv8 <a href="#userland-assembly">userland</a> and <a href="#baremetal-setup">baremetal</a> assembly, <a href="#c">ANSI C</a>, <a href="#cpp">C++</a> and <a href="#posix">POSIX</a>. <a href="#gdb">GDB step debug</a> and <a href="#kgdb">KGDB</a> just work. Powered by <a href="#about-the-qemu-buildroot-setup">Buildroot</a> and <a href="#about-the-baremetal-setup">crosstool-NG</a>. Highly automated. Thoroughly documented. Automated <a href="#test-this-repo">tests</a>. "Tested" in an Ubuntu 20.04 host.</p>
</div> </div>
<div class="paragraph"> <div class="paragraph">
<p>TL;DR: <a href="#qemu-buildroot-setup-getting-started">Section 1.2.1, &#8220;QEMU Buildroot setup getting started&#8221;</a></p> <p>TL;DR: <a href="#qemu-buildroot-setup-getting-started">Section 1.2.1, &#8220;QEMU Buildroot setup getting started&#8221;</a></p>
@@ -1598,16 +1598,7 @@ pre{ white-space:pre }
</li> </li>
<li><a href="#benchmarks">26.9. Benchmarks</a> <li><a href="#benchmarks">26.9. Benchmarks</a>
<ul class="sectlevel3"> <ul class="sectlevel3">
<li><a href="#parsec-benchmark">26.9.1. PARSEC benchmark</a> <li><a href="#microbenchmarks">26.9.1. Microbenchmarks</a></li>
<ul class="sectlevel4">
<li><a href="#parsec-benchmark-without-parsecmgmt">26.9.1.1. PARSEC benchmark without parsecmgmt</a></li>
<li><a href="#parsec-change-the-input-size">26.9.1.2. PARSEC change the input size</a></li>
<li><a href="#parsec-benchmark-with-parsecmgmt">26.9.1.3. PARSEC benchmark with parsecmgmt</a></li>
<li><a href="#parsec-uninstall">26.9.1.4. PARSEC uninstall</a></li>
<li><a href="#parsec-benchmark-hacking">26.9.1.5. PARSEC benchmark hacking</a></li>
</ul>
</li>
<li><a href="#microbenchmarks">26.9.2. Microbenchmarks</a></li>
</ul> </ul>
</li> </li>
<li><a href="#userland-libs-directory">26.10. userland/libs directory</a> <li><a href="#userland-libs-directory">26.10. userland/libs directory</a>
@@ -3045,6 +3036,19 @@ j = 0</pre>
<div class="sect3"> <div class="sect3">
<h4 id="about-the-qemu-buildroot-setup"><a class="anchor" href="#about-the-qemu-buildroot-setup"></a><a class="link" href="#about-the-qemu-buildroot-setup">1.2.3. About the QEMU Buildroot setup</a></h4> <h4 id="about-the-qemu-buildroot-setup"><a class="anchor" href="#about-the-qemu-buildroot-setup"></a><a class="link" href="#about-the-qemu-buildroot-setup">1.2.3. About the QEMU Buildroot setup</a></h4>
<div class="paragraph"> <div class="paragraph">
<p>What QEMU and Buildroot are:</p>
</div>
<div class="ulist">
<ul>
<li>
<p><a href="#introduction-to-buildroot">Introduction to Buildroot</a></p>
</li>
<li>
<p><a href="#introduction-to-qemu">Introduction to QEMU</a></p>
</li>
</ul>
</div>
<div class="paragraph">
<p>This is our reference setup, and the best supported one, use it unless you have good reason not to.</p> <p>This is our reference setup, and the best supported one, use it unless you have good reason not to.</p>
</div> </div>
<div class="paragraph"> <div class="paragraph">
@@ -4100,7 +4104,17 @@ cd userland
<p>This setup allows you to make a tiny OS and that runs just a few instructions, use it to fully control the CPU to better understand the simulators for example, or develop your own OS if you are into that.</p> <p>This setup allows you to make a tiny OS and that runs just a few instructions, use it to fully control the CPU to better understand the simulators for example, or develop your own OS if you are into that.</p>
</div> </div>
<div class="paragraph"> <div class="paragraph">
<p>You can also use C and a subset of the C standard library because we enable <a href="https://en.wikipedia.org/wiki/Newlib">Newlib</a> by default. See also: <a href="https://electronics.stackexchange.com/questions/223929/c-standard-libraries-on-bare-metal/400077#400077" class="bare">https://electronics.stackexchange.com/questions/223929/c-standard-libraries-on-bare-metal/400077#400077</a></p> <p>You can also use C and a subset of the C standard library because we enable <a href="https://en.wikipedia.org/wiki/Newlib">Newlib</a> by default. See also:</p>
</div>
<div class="ulist">
<ul>
<li>
<p><a href="https://electronics.stackexchange.com/questions/223929/c-standard-libraries-on-bare-metal/400077#400077" class="bare">https://electronics.stackexchange.com/questions/223929/c-standard-libraries-on-bare-metal/400077#400077</a></p>
</li>
<li>
<p><a href="https://stackoverflow.com/questions/13063055/does-a-libc-os-exist/59771531#59771531" class="bare">https://stackoverflow.com/questions/13063055/does-a-libc-os-exist/59771531#59771531</a></p>
</li>
</ul>
</div> </div>
<div class="paragraph"> <div class="paragraph">
<p>Our C bare-metal compiler is built with <a href="https://github.com/crosstool-ng/crosstool-ng">crosstool-NG</a>. If you have already built <a href="#qemu-buildroot-setup">Buildroot</a> previously, you will end up with two GCCs installed. Unfortunately I don&#8217;t see a solution for this, since we need separate toolchains for Newlib on baremetal and glibc on Linux: <a href="https://stackoverflow.com/questions/38956680/difference-between-arm-none-eabi-and-arm-linux-gnueabi/38989869#38989869" class="bare">https://stackoverflow.com/questions/38956680/difference-between-arm-none-eabi-and-arm-linux-gnueabi/38989869#38989869</a></p> <p>Our C bare-metal compiler is built with <a href="https://github.com/crosstool-ng/crosstool-ng">crosstool-NG</a>. If you have already built <a href="#qemu-buildroot-setup">Buildroot</a> previously, you will end up with two GCCs installed. Unfortunately I don&#8217;t see a solution for this, since we need separate toolchains for Newlib on baremetal and glibc on Linux: <a href="https://stackoverflow.com/questions/38956680/difference-between-arm-none-eabi-and-arm-linux-gnueabi/38989869#38989869" class="bare">https://stackoverflow.com/questions/38956680/difference-between-arm-none-eabi-and-arm-linux-gnueabi/38989869#38989869</a></p>
@@ -29988,7 +30002,7 @@ gensim/models/armv8/isa.ac
<div class="ulist"> <div class="ulist">
<ul> <ul>
<li> <li>
<p>the first build takes a while, but it is well worth it</p> <p>the first build takes a while compared to downloading prebuilts, but it is well worth it</p>
</li> </li>
<li> <li>
<p>the selection of software packages is relatively limited if compared to Debian.</p> <p>the selection of software packages is relatively limited if compared to Debian.</p>
@@ -30008,6 +30022,9 @@ gensim/models/armv8/isa.ac
<p>This repo basically wraps around that, and tries to make everything even more awesome for kernel developers by adding the capability of seamlessly running the stuff you&#8217;ve built on emulators usually via <code>./run</code>.</p> <p>This repo basically wraps around that, and tries to make everything even more awesome for kernel developers by adding the capability of seamlessly running the stuff you&#8217;ve built on emulators usually via <code>./run</code>.</p>
</div> </div>
<div class="paragraph"> <div class="paragraph">
<p>This runnable part of selecting the command line options for different emulators and setups is to a large extent what <a href="https://en.wikipedia.org/wiki/Libvirt">Libvirt</a> does. But we feel that having both build and run on the same repository is the key.</p>
</div>
<div class="paragraph">
<p>As this repo develops however, we&#8217;ve started taking some of the build out of Buildroot, e.g. notably the <a href="#buildroot-vanilla-kernel">Linux kernel</a> to have more build flexibility and faster build startup times.</p> <p>As this repo develops however, we&#8217;ve started taking some of the build out of Buildroot, e.g. notably the <a href="#buildroot-vanilla-kernel">Linux kernel</a> to have more build flexibility and faster build startup times.</p>
</div> </div>
<div class="paragraph"> <div class="paragraph">
@@ -30317,9 +30334,6 @@ make menuconfig</pre>
<div class="paragraph"> <div class="paragraph">
<p>Also mentioned at: <a href="https://stackoverflow.com/questions/47320800/how-to-clean-only-target-in-buildroot" class="bare">https://stackoverflow.com/questions/47320800/how-to-clean-only-target-in-buildroot</a></p> <p>Also mentioned at: <a href="https://stackoverflow.com/questions/47320800/how-to-clean-only-target-in-buildroot" class="bare">https://stackoverflow.com/questions/47320800/how-to-clean-only-target-in-buildroot</a></p>
</div> </div>
<div class="paragraph">
<p>See this for a sample manual workaround: <a href="#parsec-uninstall">Section 26.9.1.4, &#8220;PARSEC uninstall&#8221;</a>.</p>
</div>
</div> </div>
<div class="sect2"> <div class="sect2">
<h3 id="br2-target-rootfs-ext2-size"><a class="anchor" href="#br2-target-rootfs-ext2-size"></a><a class="link" href="#br2-target-rootfs-ext2-size">25.7. BR2_TARGET_ROOTFS_EXT2_SIZE</a></h3> <h3 id="br2-target-rootfs-ext2-size"><a class="anchor" href="#br2-target-rootfs-ext2-size"></a><a class="link" href="#br2-target-rootfs-ext2-size">25.7. BR2_TARGET_ROOTFS_EXT2_SIZE</a></h3>
@@ -33119,257 +33133,7 @@ cblas_dgemm( CblasColMajor, CblasNoTrans, CblasTrans,3,3,2 ,1, A,3, B,
</ul> </ul>
</div> </div>
<div class="sect3"> <div class="sect3">
<h4 id="parsec-benchmark"><a class="anchor" href="#parsec-benchmark"></a><a class="link" href="#parsec-benchmark">26.9.1. PARSEC benchmark</a></h4> <h4 id="microbenchmarks"><a class="anchor" href="#microbenchmarks"></a><a class="link" href="#microbenchmarks">26.9.1. Microbenchmarks</a></h4>
<div class="paragraph">
<p>We have ported parts of the <a href="http://parsec.cs.princeton.edu">PARSEC benchmark</a> for cross compilation at: <a href="https://github.com/cirosantilli/parsec-benchmark" class="bare">https://github.com/cirosantilli/parsec-benchmark</a> See the documentation on that repo to find out which benchmarks have been ported. Some of the benchmarks were are segfaulting, they are documented in that repo.</p>
</div>
<div class="paragraph">
<p>There are two ways to run PARSEC with this repo:</p>
</div>
<div class="ulist">
<ul>
<li>
<p><a href="#parsec-benchmark-without-parsecmgmt">without <code>pasecmgmt</code></a>, most likely what you want</p>
</li>
<li>
<p><a href="#parsec-benchmark-with-parsecmgmt">with <code>pasecmgmt</code></a></p>
</li>
</ul>
</div>
<div class="sect4">
<h5 id="parsec-benchmark-without-parsecmgmt"><a class="anchor" href="#parsec-benchmark-without-parsecmgmt"></a><a class="link" href="#parsec-benchmark-without-parsecmgmt">26.9.1.1. PARSEC benchmark without parsecmgmt</a></h5>
<div class="literalblock">
<div class="content">
<pre>./build --arch arm --download-dependencies gem5-buildroot parsec-benchmark
./build-buildroot --arch arm --config 'BR2_PACKAGE_PARSEC_BENCHMARK=y'
./run --arch arm --emulator gem5</pre>
</div>
</div>
<div class="paragraph">
<p>Once inside the guest, launch one of the <code>test</code> input sized benchmarks manually as in:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>cd /parsec/ext/splash2x/apps/fmm/run
../inst/arm-linux.gcc/bin/fmm 1 &lt; input_1</pre>
</div>
</div>
<div class="paragraph">
<p>To find run out how to run many of the benchmarks, have a look at the <code>test.sh</code> script of the <code>parse-benchmark</code> repo.</p>
</div>
<div class="paragraph">
<p>From the guest, you can also run it as:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>cd /parsec
./test.sh</pre>
</div>
</div>
<div class="paragraph">
<p>but this might be a bit time consuming in gem5.</p>
</div>
</div>
<div class="sect4">
<h5 id="parsec-change-the-input-size"><a class="anchor" href="#parsec-change-the-input-size"></a><a class="link" href="#parsec-change-the-input-size">26.9.1.2. PARSEC change the input size</a></h5>
<div class="paragraph">
<p>Running a benchmark of a size different than <code>test</code>, e.g. <code>simsmall</code>, requires a rebuild with:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>./build-buildroot \
--arch arm \
--config 'BR2_PACKAGE_PARSEC_BENCHMARK=y' \
--config 'BR2_PACKAGE_PARSEC_BENCHMARK_INPUT_SIZE="simsmall"' \
-- parsec_benchmark-reconfigure \
;</pre>
</div>
</div>
<div class="paragraph">
<p>Large input may also require tweaking:</p>
</div>
<div class="ulist">
<ul>
<li>
<p><a href="#br2-target-rootfs-ext2-size">BR2_TARGET_ROOTFS_EXT2_SIZE</a> if the unpacked inputs are large</p>
</li>
<li>
<p><a href="#memory-size">Memory size</a>, unless you want to meet the OOM killer, which is admittedly kind of fun</p>
</li>
</ul>
</div>
<div class="paragraph">
<p><code>test.sh</code> only contains the run commands for the <code>test</code> size, and cannot be used for <code>simsmall</code>.</p>
</div>
<div class="paragraph">
<p>The easiest thing to do, is to <a href="https://superuser.com/questions/231002/how-can-i-search-within-the-output-buffer-of-a-tmux-shell/1253137#1253137">scroll up on the host shell</a> after the build, and look for a line of type:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>Running /root/linux-kernel-module-cheat/out/aarch64/buildroot/build/parsec-benchmark-custom/ext/splash2x/apps/ocean_ncp/inst/aarch64-linux.gcc/bin/ocean_ncp -n2050 -p1 -e1e-07 -r20000 -t28800</pre>
</div>
</div>
<div class="paragraph">
<p>and then tweak the command found in <code>test.sh</code> accordingly.</p>
</div>
<div class="paragraph">
<p>Yes, we do run the benchmarks on host just to unpack / generate inputs. They are expected fail to run since they were build for the guest instead of host, including for x86_64 guest which has a different interpreter than the host&#8217;s (see <code>file myexecutable</code>).</p>
</div>
<div class="paragraph">
<p>The rebuild is required because we unpack input files on the host.</p>
</div>
<div class="paragraph">
<p>Separating input sizes also allows to create smaller images when only running the smaller benchmarks.</p>
</div>
<div class="paragraph">
<p>This limitation exists because <code>parsecmgmt</code> generates the input files just before running via the Bash scripts, but we can&#8217;t run <code>parsecmgmt</code> on gem5 as it is too slow!</p>
</div>
<div class="paragraph">
<p>One option would be to do that inside the guest with QEMU.</p>
</div>
<div class="paragraph">
<p>Also, we can&#8217;t generate all input sizes at once, because many of them have the same name and would overwrite one another&#8230;&#8203;</p>
</div>
<div class="paragraph">
<p>PARSEC simply wasn&#8217;t designed with non native machines in mind&#8230;&#8203;</p>
</div>
</div>
<div class="sect4">
<h5 id="parsec-benchmark-with-parsecmgmt"><a class="anchor" href="#parsec-benchmark-with-parsecmgmt"></a><a class="link" href="#parsec-benchmark-with-parsecmgmt">26.9.1.3. PARSEC benchmark with parsecmgmt</a></h5>
<div class="paragraph">
<p>Most users won&#8217;t want to use this method because:</p>
</div>
<div class="ulist">
<ul>
<li>
<p>running the <code>parsecmgmt</code> Bash scripts takes forever before it ever starts running the actual benchmarks on gem5</p>
<div class="paragraph">
<p>Running on QEMU is feasible, but not the main use case, since QEMU cannot be used for performance measurements</p>
</div>
</li>
<li>
<p>it requires putting the full <code>.tar</code> inputs on the guest, which makes the image twice as large (1x for the <code>.tar</code>, 1x for the unpacked input files)</p>
</li>
</ul>
</div>
<div class="paragraph">
<p>It would be awesome if it were possible to use this method, since this is what Parsec supports officially, and so:</p>
</div>
<div class="ulist">
<ul>
<li>
<p>you don&#8217;t have to dig into what raw command to run</p>
</li>
<li>
<p>there is an easy way to run all the benchmarks in one go to test them out</p>
</li>
<li>
<p>you can just run any of the benchmarks that you want</p>
</li>
</ul>
</div>
<div class="paragraph">
<p>but it simply is not feasible in gem5 because it takes too long.</p>
</div>
<div class="paragraph">
<p>If you still want to run this, try it out with:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>./build-buildroot \
--arch aarch64 \
--config 'BR2_PACKAGE_PARSEC_BENCHMARK=y' \
--config 'BR2_PACKAGE_PARSEC_BENCHMARK_PARSECMGMT=y' \
--config 'BR2_TARGET_ROOTFS_EXT2_SIZE="3G"' \
-- parsec_benchmark-reconfigure \
;</pre>
</div>
</div>
<div class="paragraph">
<p>And then you can run it just as you would on the host:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>cd /parsec/
bash
. env.sh
parsecmgmt -a run -p splash2x.fmm -i test</pre>
</div>
</div>
</div>
<div class="sect4">
<h5 id="parsec-uninstall"><a class="anchor" href="#parsec-uninstall"></a><a class="link" href="#parsec-uninstall">26.9.1.4. PARSEC uninstall</a></h5>
<div class="paragraph">
<p>If you want to remove PARSEC later, Buildroot doesn&#8217;t provide an automated package removal mechanism as mentioned at: <a href="#remove-buildroot-packages">Section 25.6, &#8220;Remove Buildroot packages&#8221;</a>, but the following procedure should be satisfactory:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>rm -rf \
"$(./getvar buildroot_download_dir)"/parsec-* \
"$(./getvar buildroot_build_dir)"/build/parsec-* \
"$(./getvar buildroot_build_dir)"/build/packages-file-list.txt \
"$(./getvar buildroot_build_dir)"/images/rootfs.* \
"$(./getvar buildroot_build_dir)"/target/parsec-* \
;
./build-buildroot --arch arm</pre>
</div>
</div>
</div>
<div class="sect4">
<h5 id="parsec-benchmark-hacking"><a class="anchor" href="#parsec-benchmark-hacking"></a><a class="link" href="#parsec-benchmark-hacking">26.9.1.5. PARSEC benchmark hacking</a></h5>
<div class="paragraph">
<p>If you end up going inside <a href="https://github.com/cirosantilli/linux-kernel-module-cheat/blob/master/submodules/parsec-benchmark">submodules/parsec-benchmark</a> to hack up the benchmark (you will!), these tips will be helpful.</p>
</div>
<div class="paragraph">
<p>Buildroot was not designed to deal with large images, and currently cross rebuilds are a bit slow, due to some image generation and validation steps.</p>
</div>
<div class="paragraph">
<p>A few workarounds are:</p>
</div>
<div class="ulist">
<ul>
<li>
<p>develop in host first as much as you can. Our PARSEC fork supports it.</p>
<div class="paragraph">
<p>If you do this, don&#8217;t forget to do a:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>cd "$(./getvar parsec_source_dir)"
git clean -xdf .</pre>
</div>
</div>
<div class="paragraph">
<p>before going for the cross compile build.</p>
</div>
</li>
<li>
<p>patch Buildroot to work well, and keep cross compiling all the way. This should be totally viable, and we should do it.</p>
<div class="paragraph">
<p>Don&#8217;t forget to explicitly rebuild PARSEC with:</p>
</div>
<div class="literalblock">
<div class="content">
<pre>./build-buildroot \
--arch arm \
--config 'BR2_PACKAGE_PARSEC_BENCHMARK=y' \
-- parsec_benchmark-reconfigure \
;</pre>
</div>
</div>
<div class="paragraph">
<p>You may also want to test if your patches are still functionally correct inside of QEMU first, which is a faster emulator.</p>
</div>
</li>
<li>
<p>sell your soul, and compile natively inside the guest. We won&#8217;t do this, not only because it is evil, but also because Buildroot explicitly does not support it: <a href="https://buildroot.org/downloads/manual/manual.html#faq-no-compiler-on-target" class="bare">https://buildroot.org/downloads/manual/manual.html#faq-no-compiler-on-target</a> ARM employees have been known to do this: <a href="https://github.com/arm-university/arm-gem5-rsk/blob/aa3b51b175a0f3b6e75c9c856092ae0c8f2a7cdc/parsec_patches/qemu-patch.diff" class="bare">https://github.com/arm-university/arm-gem5-rsk/blob/aa3b51b175a0f3b6e75c9c856092ae0c8f2a7cdc/parsec_patches/qemu-patch.diff</a></p>
</li>
</ul>
</div>
</div>
</div>
<div class="sect3">
<h4 id="microbenchmarks"><a class="anchor" href="#microbenchmarks"></a><a class="link" href="#microbenchmarks">26.9.2. Microbenchmarks</a></h4>
<div class="paragraph"> <div class="paragraph">
<p>It eventually has to come to that, hasn&#8217;t it?</p> <p>It eventually has to come to that, hasn&#8217;t it?</p>
</div> </div>