Re: [Qemu-devel] [Qemu-ppc] macio ide question/bug report

[BALATON Zoltan <balaton@eik.bme.hu>]

On Thu, 15 May 2014, BALATON Zoltan wrote:
> confusing.) Do you think that replacing io->len in your patch with 
> s->io_buffer_size would be the correct thing to do?

Probably that's not enough. I've tried it and then it gets to here:

end of non-IO ATAPI DMA transfer
122	    if (!s->packet_transfer_size) {
(gdb) p s->packet_transfer_size
$1 = 0
(gdb) n
123	        MACIO_DPRINTF("end of transfer\n");
end of transfer
124	        ide_atapi_cmd_ok(s);
125	        m->dma_active = false;
128	    if (io->len == 0) {
(gdb) p io->len
$2 = 804
(gdb) l
123	        MACIO_DPRINTF("end of transfer\n");
124	        ide_atapi_cmd_ok(s);
125	        m->dma_active = false;
126	    }
127
128	    if (io->len == 0) {
129	        MACIO_DPRINTF("end of DMA\n");
130	        goto done;
131	    }
132
133	    /* launch next transfer */
134
135	    /* handle unaligned accesses first, get them over with and only do the
136	       remaining bulk transfer using our async DMA helpers */
137	    unaligned = io->len & 0x1ff;
138	    if (unaligned) {
139	        int sector_num = (s->lba << 2) + (s->io_buffer_index >> 9);
140	        int nsector = io->len >> 9;
141
142	        MACIO_DPRINTF("precopying unaligned %d bytes to %#" HWADDR_PRIx "\n",
(gdb) c
Continuing.
precopying unaligned 292 bytes to 0xe4f50c
io->len = 0x200
set remainder to: -20
sector_num=-4 size=0, cmd_cmd=0
atapi_cmd_error: sense=0x5 asc=0x21
done DMA
DBDMA: dbdma_end

Regards,
BALATON Zoltan