diff options
| author | Christian Krinitsin <mail@krinitsin.com> | 2025-07-03 07:27:52 +0000 |
|---|---|---|
| committer | Christian Krinitsin <mail@krinitsin.com> | 2025-07-03 07:27:52 +0000 |
| commit | d0c85e36e4de67af628d54e9ab577cc3fad7796a (patch) | |
| tree | f8f784b0f04343b90516a338d6df81df3a85dfa2 /results/classifier/gemma3:12b/kernel/2855 | |
| parent | 7f4364274750eb8cb39a3e7493132fca1c01232e (diff) | |
| download | qemu-analysis-d0c85e36e4de67af628d54e9ab577cc3fad7796a.tar.gz qemu-analysis-d0c85e36e4de67af628d54e9ab577cc3fad7796a.zip | |
add deepseek and gemma results
Diffstat (limited to 'results/classifier/gemma3:12b/kernel/2855')
| -rw-r--r-- | results/classifier/gemma3:12b/kernel/2855 | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/results/classifier/gemma3:12b/kernel/2855 b/results/classifier/gemma3:12b/kernel/2855 new file mode 100644 index 000000000..1ebe5d467 --- /dev/null +++ b/results/classifier/gemma3:12b/kernel/2855 @@ -0,0 +1,30 @@ + +masking mode field in mepc before mret +Description of problem: +I thought I found a bug in OpenSBI (https://github.com/riscv-software-src/opensbi/issues/391) but it actually is a QEMU bug. +It is described here: https://lists.infradead.org/pipermail/opensbi/2025-March/008166.html +Steps to reproduce: +1. use an application with vectored mode enabled (The RISC-V Instruction Set Manual: Volume II: Privileged Architecture / chapter 10.1.2) in QEMU +2. trigger an illegal instruction interrupt (handle it in machine mode - not by medeleg) +3. in a machine mode trap: Store STVEC in MEPC. +4. do a mret +5. the first bits of mepc are not masked so the address in mepc (comming from (v)stvec) will be false after mret +Additional information: +My guess is that the instructions from the following quote (masking of lower bits in mepc) from the official spec must be implemented here: +https://gitlab.com/qemu-project/qemu/-/blob/master/target/riscv/op_helper.c?ref_type=heads#L387 +Maybe also somewhere else. + +> 3.1.14. Machine Exception Program Counter (mepc) +> +> mepc is an MXLEN-bit read/write register formatted as shown in Figure 21. The low bit of mepc +> (mepc[0]) is always zero. On implementations that support only IALIGN=32, the two low bits +> (mepc[1:0]) are always zero. +> +> If an implementation allows IALIGN to be either 16 or 32 (by changing CSR misa, for example), then, +> whenever IALIGN=32, bit mepc[1] is masked on reads so that it appears to be 0. This masking occurs +> also for the implicit read by the MRET instruction. Though masked, mepc[1] remains writable when +> IALIGN=32. +> +> mepc is a WARL register that must be able to hold all valid virtual addresses. It need not be capable of +> holding all possible invalid addresses. Prior to writing mepc, implementations may convert an invalid +> address into some other invalid address that mepc is capable of holding. |