On 15 November 2018 at 19:24, <miny...@acm.org> wrote: > From: Corey Minyard <cminy...@mvista.com> > > It is never supposed to fail and cannot return an error, so just > have it return the proper type. Have it return 0xff on nothing > available, since that's what would happen on a real bus. > > Signed-off-by: Corey Minyard <cminy...@mvista.com>
This is a nice cleanup. There are some followups at callsites of i2c_recv(): hw/arm/stellaris.c: s->mdr = i2c_recv(s->bus) & 0xff; -- mask not needed hw/i2c/aspeed_i2c.c: ret = i2c_recv(bus->bus); hw/i2c/aspeed_i2c.c- if (ret < 0) { hw/i2c/aspeed_i2c.c- qemu_log_mask(LOG_GUEST_ERROR, "%s: read failed\n", __func__); hw/i2c/aspeed_i2c.c- ret = 0xff; -- error handling can be deleted hw/i2c/exynos4210_i2c.c: ret = i2c_recv(s->bus); hw/i2c/exynos4210_i2c.c- if (ret < 0 && (s->i2ccon & I2CCON_ACK_GEN)) { hw/i2c/exynos4210_i2c.c- s->i2cstat |= I2CSTAT_LAST_BIT; /* Data is not acknowledged */ -- ditto hw/i2c/imx_i2c.c: ret = i2c_recv(s->bus); hw/i2c/imx_i2c.c- hw/i2c/imx_i2c.c- if (ret >= 0) { hw/i2c/imx_i2c.c- imx_i2c_raise_interrupt(s); -- condition now always-true which should probably go in their own patch(es). Reviewed-by: Peter Maydell <peter.mayd...@linaro.org> thanks -- PMM