Re: [Qemu-devel] [PATCH v2 20/22] tcg: Save insn data and use it in cpu_restore_state_from_tb

[Richard Henderson <rth@twiddle.net>]

On 09/18/2015 06:08 AM, Peter Maydell wrote:
> On 18 September 2015 at 05:55, Richard Henderson <rth@twiddle.net> wrote:
>> We can now restore state without retranslation.
>>
>> Signed-off-by: Richard Henderson <rth@twiddle.net>
>> ---
>> +/* Encode the data collected about the instructions while compiling TB.
>> +   Place the data at BLOCK, and return the number of bytes consumed.
>> +
>> +   The logical table consisits of TARGET_INSN_START_WORDS target_ulong's,
>
> "consists". No apostrophe in 'target_ulongs'.
>
>> +   which come from the target's insn_start data, followed by a uintptr_t
>> +   which comes from the host pc of the end of the code implementing the insn.
>> +
>> +   Each line of the table is encoded as sleb128 deltas from the previous
>> +   line.  The seed for the first line is { tb->pc, 0..., tb->tc_ptr }.
>> +   That is, the first column is seeded with the guest pc, the last column
>> +   with the host pc, and the middle columns with zeros.  */
>
> You're still not allowing for your worst-case datatable size when we
> calculate tcg_ctx.code_gen_buffer_max_size.

I'll note that the current worst-case estimate is way too big: 122kB.

Which honestly means we're wasting a ton of space at the end of the 
code_gen_buffer.  While down-thread we talk about guard pages and sigsegv 
handlers etc, I now believe this shouldn't be a blocker for this patch set.

(And in particular, setting up a SEH handler for Win32 to act as a sigsegv 
handler is just too annoyingly difficult.  It'd be one thing if we only 
targeted VC++, but doing SEH in GCC at present is just Too Ugly.  So we'd have 
two different schemes for win32 and posix, which doesn't seem to be the best of 
ideas.)


r~