Re: [Qemu-trivial] [PATCH v2 2/7] hw/mips/mips_jazz.c: Store irq array in MachineState to fix memory leak

[Shannon Zhao]

On 2015/6/4 22:18, Michael Tokarev wrote:
> 30.05.2015 10:54, Shannon Zhao пишет:
> > >From: Shannon Zhao<address@hidden>
> > >
> > >Signed-off-by: Shannon Zhao<address@hidden>
> > >Signed-off-by: Shannon Zhao<address@hidden>
> > >---
> > >  hw/mips/mips_jazz.c | 28 ++++++++++++++--------------
> > >  1 file changed, 14 insertions(+), 14 deletions(-)
> > >
> > >diff --git a/hw/mips/mips_jazz.c b/hw/mips/mips_jazz.c
> > >index 2c153e0..259458b 100644
> > >--- a/hw/mips/mips_jazz.c
> > >+++ b/hw/mips/mips_jazz.c
> > >@@ -135,7 +135,7 @@ static void mips_jazz_init(MachineState *machine,
> > >      MIPSCPU *cpu;
> > >      CPUClass *cc;
> > >      CPUMIPSState *env;
> > >-    qemu_irq *rc4030, *i8259;
> > >+    qemu_irq *i8259;
> Hm. Why do you only cover rc4030, not i8259?

As i8259 is stored in ISABus->irqs by function isa_bus_irqs.

void isa_bus_irqs(ISABus *bus, qemu_irq *irqs)
{
    if (!bus) {
        hw_error("Can't set isa irqs with no isa bus present.");
    }
    bus->irqs = irqs;
}

> Besides, in order to keep the changes smaller, I think it is okay to
> keep the variables like that, here and in the rest of the function,
> and only add assignment of it to machine->irqs.  This way, we also
> keep semantic names of the variables, rc4030[i] is easier to understand
> than machine->irqs[i], the former's more specific.
Agree.

> BTW, there's also cpu_exit_irq in this function whose allocation also
> suffers from qemu_allocate_irqs(..., 1) API abuse.

Yeah, missed this one.

--
Shannon