summary refs log tree commit diff stats
path: root/results/classifier/semantic-bugs/instruction/1915027
diff options
context:
space:
mode:
authorChristian Krinitsin <mail@krinitsin.com>2025-06-12 09:56:59 +0200
committerChristian Krinitsin <mail@krinitsin.com>2025-06-12 09:56:59 +0200
commitb89a938452613061c0f1f23e710281cf5c83cb29 (patch)
treed5faecfd167e088848cad894f8dc9cfef3352e3b /results/classifier/semantic-bugs/instruction/1915027
parent7b681b9f9eedaad2f081ae11a32f459f5a1312ff (diff)
downloadqemu-analysis-b89a938452613061c0f1f23e710281cf5c83cb29.tar.gz
qemu-analysis-b89a938452613061c0f1f23e710281cf5c83cb29.zip
add manually reviewed semantic bugs
Diffstat (limited to 'results/classifier/semantic-bugs/instruction/1915027')
-rw-r--r--results/classifier/semantic-bugs/instruction/191502727
1 files changed, 27 insertions, 0 deletions
diff --git a/results/classifier/semantic-bugs/instruction/1915027 b/results/classifier/semantic-bugs/instruction/1915027
new file mode 100644
index 000000000..4df7ceb4b
--- /dev/null
+++ b/results/classifier/semantic-bugs/instruction/1915027
@@ -0,0 +1,27 @@
+instruction: 0.816
+assembly: 0.781
+graphic: 0.759
+other: 0.744
+device: 0.634
+semantic: 0.632
+mistranslation: 0.442
+vnc: 0.284
+network: 0.181
+boot: 0.140
+socket: 0.116
+KVM: 0.044
+
+RISC-V 64, CPUs do ilegal 0x00 write with SMP
+
+When QEMU is runt like this:
+
+qemu-system-riscv64 -d unimp,guest_errors -smp 8
+
+Other harts will do a illegal write on address 0x00.
+
+This could be mostly (i think) because the initial assembly code is only loaded on the first hart and the others do a mess because there is no code to execute.
+
+Even with -smp 1 you will see the same errors. The problem is because there is nothing to run after OpenSBI jumps to the next stage.
+
+If you load a kernel you will not see the error messages.
+