On Sat, Mar 31, 2012 at 09:19:42PM +0800, Li Zhi Hui wrote:
> Replace bdrv_* to bdrv_aio_* functions in DMA mode in fdc.c.
>
> Signed-off-by: Li Zhi Hui <zhihu...@linux.vnet.ibm.com>
> ---
> hw/dma.c | 36 +++++----
> hw/fdc.c | 260
> +++++++++++++++++++++++++++++++++++++++++++++-----------------
> hw/isa.h | 1 +
> 3 files changed, 212 insertions(+), 85 deletions(-)
>
> diff --git a/hw/dma.c b/hw/dma.c
> index 0a9322d..8f4275b 100644
> --- a/hw/dma.c
> +++ b/hw/dma.c
> @@ -357,15 +357,6 @@ static void DMA_run (void)
> {
> struct dma_cont *d;
> int icont, ichan;
> - int rearm = 0;
> - static int running = 0;
> -
> - if (running) {
> - rearm = 1;
> - goto out;
> - } else {
> - running = 1;
> - }
>
> d = dma_controllers;
>
> @@ -377,15 +368,9 @@ static void DMA_run (void)
>
> if ((0 == (d->mask & mask)) && (0 != (d->status & (mask << 4))))
> {
> channel_run (icont, ichan);
> - rearm = 1;
> }
> }
> }
> -
> - running = 0;
> -out:
> - if (rearm)
> - qemu_bh_schedule_idle(dma_bh);
> }
The 'running' variable is used to protect against reentrancy. Think
about this situation: DMA_run() -> device_foo_channel_handler(icont,
ichan) -> DMA_run().
The 'running' variable ensures that we schedule the BH and return.
Since you removed it we would run more channel handlers inside the
nested DMA_run() call.
Why did you remove 'running'?
>
> static void DMA_run_bh(void *unused)
> @@ -460,6 +445,27 @@ void DMA_schedule(int nchan)
> qemu_irq_pulse(*d->cpu_request_exit);
> }
>
> +void DMA_set_return(int nret)
> +{
> + struct dma_regs *r;
> + struct dma_cont *d;
> + int icont, ichan;
> +
> + d = dma_controllers;
> + for (icont = 0; icont < 2; icont++, d++) {
> + for (ichan = 0; ichan < 4; ichan++) {
> + int mask;
> +
> + mask = 1 << ichan;
> +
> + if ((0 == (d->mask & mask)) && (0 != (d->status & (mask << 4))))
> {
> + r = &d[icont].regs[ichan];
> + r->now[COUNT] = nret;
> + }
> + }
> + }
> +}
I think multiple channels can be active at the same time. So we need a
int nchan argument to identify the channel, just like DMA_hold_DREQ() or
DMA_release_DREQ().
I suggest making the hw/dma.c changes in a separate patch in this series.
> /* handlers for DMA transfers */
> static int fdctrl_transfer_handler (void *opaque, int nchan,
> int dma_pos, int dma_len)
> @@ -1189,6 +1310,11 @@ static int fdctrl_transfer_handler (void *opaque, int
> nchan,
> FDrive *cur_drv;
> int len, start_pos, rel_pos;
> uint8_t status0 = 0x00, status1 = 0x00, status2 = 0x00;
> + int fdc_sector_num = 0;
> + uint8_t *pfdc_string = NULL;
> + FDC_opaque *opaque_cb;
> + bool write_flag = FALSE;
Please use 'true' and 'false', they are in <stdbool.h> together with the
bool type.
> + int write_sector = 0;
>
> fdctrl = opaque;
> if (fdctrl->msr & FD_MSR_RQM) {
> @@ -1196,107 +1322,101 @@ static int fdctrl_transfer_handler (void *opaque,
> int nchan,
> return 0;
> }
> cur_drv = get_cur_drv(fdctrl);
> - if (fdctrl->data_dir == FD_DIR_SCANE || fdctrl->data_dir == FD_DIR_SCANL
> ||
> - fdctrl->data_dir == FD_DIR_SCANH)
> + if ((fdctrl->data_dir == FD_DIR_SCANE) ||
> + (fdctrl->data_dir == FD_DIR_SCANL) ||
> + (fdctrl->data_dir == FD_DIR_SCANH)) {
> status2 = FD_SR2_SNS;
> - if (dma_len > fdctrl->data_len)
> + }
> + if (dma_len > fdctrl->data_len) {
> dma_len = fdctrl->data_len;
> + }
> if (cur_drv->bs == NULL) {
> - if (fdctrl->data_dir == FD_DIR_WRITE)
> - fdctrl_stop_transfer(fdctrl, FD_SR0_ABNTERM | FD_SR0_SEEK, 0x00,
> 0x00);
> - else
> + if (fdctrl->data_dir == FD_DIR_WRITE) {
> + fdctrl_stop_transfer(fdctrl,
> + FD_SR0_ABNTERM | FD_SR0_SEEK, 0x00, 0x00);
> + } else {
> fdctrl_stop_transfer(fdctrl, FD_SR0_ABNTERM, 0x00, 0x00);
> + }
> len = 0;
> goto transfer_error;
> }
> +
> + if ((fdctrl->data_dir != FD_DIR_WRITE) && (fdctrl->data_pos < dma_len)) {
> + fdc_sector_num = (dma_len + FD_SECTOR_LEN - 1) / FD_SECTOR_LEN;
> + opaque_cb = g_malloc0(sizeof(FDC_opaque));
> + pfdc_string = g_malloc0(fdc_sector_num * FD_SECTOR_LEN);
bdrv_*() I/O buffers should be allocated with qemu_blockalign() and
freed with qemu_vfree().
> + opaque_cb->fdctrl = fdctrl;
> + opaque_cb->nchan = nchan;
> + opaque_cb->dma_len = dma_len;
> +
> + fdctrl->iov.iov_base = pfdc_string;
> + fdctrl->iov.iov_len = fdc_sector_num * FD_SECTOR_LEN;
> + qemu_iovec_init_external(&fdctrl->qiov, &fdctrl->iov, 1);
> + bdrv_aio_readv(cur_drv->bs, fd_sector(cur_drv),
> + &fdctrl->qiov, fdc_sector_num, fdctrl_read_DMA_cb, opaque_cb);
> + return dma_len;
Should be return 0 since we haven't completed I/O yet.
Stefan
