HI,
I try to understand the detail process of read a block in domU , but meet some problem, confused me.
first, i change funcntion void do_blkif_request(request_queue_t *rq), from
DPRINTK("do_blk_req %p: cmd %p, sec %llx, "
"(%u/%li) buffer:%p [%s]\n",
req, req->cmd, (long long)req->sector,
req->current_nr_sectors,
req->nr_sectors, req->buffer,
rq_data_dir(req) ? "write" : "read"); to
printk("do_blk_req %p: cmd %p, sec %llx, "
"(%u/%li) buffer:%p [%s]\n",
req, req->cmd, (long long)req->sector,
req->current_nr_sectors,
req->nr_sectors, req->buffer,
rq_data_dir(req) ? "write" : "read"); int order to see the message. when domU read blocks.
when boot up domU , some message displayed as follow:
do_blk_req cf9d1d3c: cmd cf9d1db8, sec 6d032, (2/2) buffer:cfae5000 [read]
do_blk_req cf9d1d3c: cmd cf9d1db8, sec 6c2d2, (2/2) buffer:d07b3400 [read] do_blk_req cf9d1d3c: cmd cf9d1db8, sec 6d034, (2/2) buffer:cfae6000 [read] do_blk_req cf9d1de8: cmd cf9d1e64, sec 6df0e, (2/2) buffer:cfae6400 [read]
do_blk_req cf9d1f40: cmd cf9d1fbc, sec 6df10, (2/2) buffer:cfae6800 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 71bec, (2/2) buffer:cfae6c00 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 71c54, (2/2) buffer:cfae7000 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 6c2d0, (2/2) buffer:d07b3000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 71ca6, (8/26) buffer:cfae8000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 71cc0, (2/2) buffer:cfafa000 [read]
do_blk_req cf9d1f40: cmd cf9d1fbc, sec 6c248, (2/2) buffer:cfaed000 [read] do_blk_req cf9d1f40: cmd cf9d1fbc, sec 6e754, (8/10) buffer:cfafb000 [read] do_blk_req cf9d1f40: cmd cf9d1fbc, sec 71c92, (6/6) buffer:cfafd000 [read]
do_blk_req cf9d1f40: cmd cf9d1fbc, sec 6c2ce, (2/2) buffer:cfb1ec00 [read] do_blk_req cf9d1f40: cmd cf9d1fbc, sec 71a0e, (2/2) buffer:cfafe000 [read] do_blk_req cf9d1f40: cmd cf9d1fbc, sec 4758e, (8/64) buffer:cfaff000 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 475ce, (8/72) buffer:cfb43000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 50016, (2/2) buffer:d0389c00 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 528d0, (2/2) buffer:cfb4c000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 642d6, (2/2) buffer:d07adc00 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 66518, (2/2) buffer:cfb4d000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 680ce, (2/2) buffer:cf416c00 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 6ad2c, (2/2) buffer:cfb4e000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 7007e, (2/2) buffer:cf418c00 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 73894, (2/2) buffer:cfb4f000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 70080, (2/2) buffer:cf419000 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 73896, (8/26) buffer:cfb50000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 738b0, (8/34) buffer:d0763000 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 84010, (2/2) buffer:cf421000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 862cc, (2/2) buffer:d0776000 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 84016, (2/2) buffer:cf421c00 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 8635e, (8/18) buffer:d0777000 [read]
do_blk_req cf9d1be4: cmd cf9d1c60, sec 86388, (6/6) buffer:cf426000 [read] do_blk_req cf9d1be4: cmd cf9d1c60, sec 6ad34, (2/2) buffer:cf427000 [read]
i don't know what req refers to , is req mean a request tag?, and req-cmd is the same mean to req?
such as the same cf9d1be4 and same cf9d1c60 above?
i guess nr_sectors == current_nr_sectors + next current_nr_sectors + next current_nr_sectors + ...
such a the yellow backgroud lines above . i think 72 equal to
8+2+2+2+2+2+2+2+2+2+8+8+2+2+2+8+6+2 == 64 , but it's not true. can you tell me why? and how to compute or get the exact num of secters when a read or write request happens.
i have see the source code segment in the request struct :
sector_t sector; /* next sector to submit */ unsigned long nr_sectors; /* no. of sectors left to submit */ /* no. of sectors left to submit in the current segment */
unsigned int current_nr_sectors; char *buffer; unsigned char cmd[BLK_MAX_CDB];
but still not very clear about the comment above.
Thanks very much !
-- Regards,
Sucan
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|