Commit abfa034e authored by Aneesh Kumar K.V's avatar Aneesh Kumar K.V Committed by Eric Van Hensbergen

fs/9p: Update zero-copy implementation in 9p

* remove lot of update to different data structure
* add a seperate callback for zero copy request.
* above makes non zero copy code path simpler
* remove conditionalizing TREAD/TREADDIR/TWRITE in the zero copy path
* Fix the dotu p9_check_errors with zero copy. Add sufficient doc around
* Add support for both in and output buffers in zero copy callback
* pin and unpin pages in the same context
* use helpers instead of defining page offset and rest of page ourself
* Fix mem leak in p9_check_errors
* Remove 'E' and 'F' in p9pdu_vwritef
Signed-off-by: default avatarAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: default avatarEric Van Hensbergen <ericvh@gmail.com>
parent c3b92c87
...@@ -231,7 +231,7 @@ static int v9fs_dir_readdir_dotl(struct file *filp, void *dirent, ...@@ -231,7 +231,7 @@ static int v9fs_dir_readdir_dotl(struct file *filp, void *dirent,
while (err == 0) { while (err == 0) {
if (rdir->tail == rdir->head) { if (rdir->tail == rdir->head) {
err = p9_client_readdir(fid, rdir->buf, buflen, err = p9_client_readdir(fid, rdir->buf, buflen,
filp->f_pos); filp->f_pos);
if (err <= 0) if (err <= 0)
goto unlock_and_exit; goto unlock_and_exit;
......
...@@ -359,6 +359,9 @@ enum p9_qid_t { ...@@ -359,6 +359,9 @@ enum p9_qid_t {
/* Room for readdir header */ /* Room for readdir header */
#define P9_READDIRHDRSZ 24 #define P9_READDIRHDRSZ 24
/* size of header for zero copy read/write */
#define P9_ZC_HDR_SZ 4096
/** /**
* struct p9_qid - file system entity information * struct p9_qid - file system entity information
* @type: 8-bit type &p9_qid_t * @type: 8-bit type &p9_qid_t
...@@ -555,10 +558,6 @@ struct p9_rstatfs { ...@@ -555,10 +558,6 @@ struct p9_rstatfs {
* @tag: transaction id of the request * @tag: transaction id of the request
* @offset: used by marshalling routines to track current position in buffer * @offset: used by marshalling routines to track current position in buffer
* @capacity: used by marshalling routines to track total malloc'd capacity * @capacity: used by marshalling routines to track total malloc'd capacity
* @pubuf: Payload user buffer given by the caller
* @pkbuf: Payload kernel buffer given by the caller
* @pbuf_size: pubuf/pkbuf(only one will be !NULL) size to be read/write.
* @private: For transport layer's use.
* @sdata: payload * @sdata: payload
* *
* &p9_fcall represents the structure for all 9P RPC * &p9_fcall represents the structure for all 9P RPC
...@@ -575,10 +574,6 @@ struct p9_fcall { ...@@ -575,10 +574,6 @@ struct p9_fcall {
size_t offset; size_t offset;
size_t capacity; size_t capacity;
char __user *pubuf;
char *pkbuf;
size_t pbuf_size;
void *private;
u8 *sdata; u8 *sdata;
}; };
......
...@@ -26,13 +26,6 @@ ...@@ -26,13 +26,6 @@
#ifndef NET_9P_TRANSPORT_H #ifndef NET_9P_TRANSPORT_H
#define NET_9P_TRANSPORT_H #define NET_9P_TRANSPORT_H
#define P9_TRANS_PREF_PAYLOAD_MASK 0x1
/* Default. Add Payload to PDU before sending it down to transport layer */
#define P9_TRANS_PREF_PAYLOAD_DEF 0x0
/* Send pay load separately to transport layer along with PDU.*/
#define P9_TRANS_PREF_PAYLOAD_SEP 0x1
/** /**
* struct p9_trans_module - transport module interface * struct p9_trans_module - transport module interface
* @list: used to maintain a list of currently available transports * @list: used to maintain a list of currently available transports
...@@ -56,13 +49,14 @@ struct p9_trans_module { ...@@ -56,13 +49,14 @@ struct p9_trans_module {
struct list_head list; struct list_head list;
char *name; /* name of transport */ char *name; /* name of transport */
int maxsize; /* max message size of transport */ int maxsize; /* max message size of transport */
int pref; /* Preferences of this transport */
int def; /* this transport should be default */ int def; /* this transport should be default */
struct module *owner; struct module *owner;
int (*create)(struct p9_client *, const char *, char *); int (*create)(struct p9_client *, const char *, char *);
void (*close) (struct p9_client *); void (*close) (struct p9_client *);
int (*request) (struct p9_client *, struct p9_req_t *req); int (*request) (struct p9_client *, struct p9_req_t *req);
int (*cancel) (struct p9_client *, struct p9_req_t *req); int (*cancel) (struct p9_client *, struct p9_req_t *req);
int (*zc_request)(struct p9_client *, struct p9_req_t *,
char *, char *, int , int, int, int);
}; };
void v9fs_register_trans(struct p9_trans_module *m); void v9fs_register_trans(struct p9_trans_module *m);
......
...@@ -203,11 +203,12 @@ static int parse_opts(char *opts, struct p9_client *clnt) ...@@ -203,11 +203,12 @@ static int parse_opts(char *opts, struct p9_client *clnt)
* *
*/ */
static struct p9_req_t *p9_tag_alloc(struct p9_client *c, u16 tag) static struct p9_req_t *p9_tag_alloc(struct p9_client *c, u16 tag, int max_size)
{ {
unsigned long flags; unsigned long flags;
int row, col; int row, col;
struct p9_req_t *req; struct p9_req_t *req;
int alloc_msize = min(c->msize, max_size);
/* This looks up the original request by tag so we know which /* This looks up the original request by tag so we know which
* buffer to read the data into */ * buffer to read the data into */
...@@ -245,23 +246,12 @@ static struct p9_req_t *p9_tag_alloc(struct p9_client *c, u16 tag) ...@@ -245,23 +246,12 @@ static struct p9_req_t *p9_tag_alloc(struct p9_client *c, u16 tag)
return ERR_PTR(-ENOMEM); return ERR_PTR(-ENOMEM);
} }
init_waitqueue_head(req->wq); init_waitqueue_head(req->wq);
if ((c->trans_mod->pref & P9_TRANS_PREF_PAYLOAD_MASK) == req->tc = kmalloc(sizeof(struct p9_fcall) + alloc_msize,
P9_TRANS_PREF_PAYLOAD_SEP) { GFP_NOFS);
int alloc_msize = min(c->msize, 4096); req->tc->capacity = alloc_msize;
req->tc = kmalloc(sizeof(struct p9_fcall)+alloc_msize, req->rc = kmalloc(sizeof(struct p9_fcall) + alloc_msize,
GFP_NOFS); GFP_NOFS);
req->tc->capacity = alloc_msize; req->rc->capacity = alloc_msize;
req->rc = kmalloc(sizeof(struct p9_fcall)+alloc_msize,
GFP_NOFS);
req->rc->capacity = alloc_msize;
} else {
req->tc = kmalloc(sizeof(struct p9_fcall)+c->msize,
GFP_NOFS);
req->tc->capacity = c->msize;
req->rc = kmalloc(sizeof(struct p9_fcall)+c->msize,
GFP_NOFS);
req->rc->capacity = c->msize;
}
if ((!req->tc) || (!req->rc)) { if ((!req->tc) || (!req->rc)) {
printk(KERN_ERR "Couldn't grow tag array\n"); printk(KERN_ERR "Couldn't grow tag array\n");
kfree(req->tc); kfree(req->tc);
...@@ -485,27 +475,8 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) ...@@ -485,27 +475,8 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req)
if (!p9_is_proto_dotl(c)) { if (!p9_is_proto_dotl(c)) {
char *ename; char *ename;
if (req->tc->pbuf_size) {
/* Handle user buffers */
size_t len = req->rc->size - req->rc->offset;
if (req->tc->pubuf) {
/* User Buffer */
err = copy_from_user(
&req->rc->sdata[req->rc->offset],
req->tc->pubuf, len);
if (err) {
err = -EFAULT;
goto out_err;
}
} else {
/* Kernel Buffer */
memmove(&req->rc->sdata[req->rc->offset],
req->tc->pkbuf, len);
}
}
err = p9pdu_readf(req->rc, c->proto_version, "s?d", err = p9pdu_readf(req->rc, c->proto_version, "s?d",
&ename, &ecode); &ename, &ecode);
if (err) if (err)
goto out_err; goto out_err;
...@@ -515,11 +486,10 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) ...@@ -515,11 +486,10 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req)
if (!err || !IS_ERR_VALUE(err)) { if (!err || !IS_ERR_VALUE(err)) {
err = p9_errstr2errno(ename, strlen(ename)); err = p9_errstr2errno(ename, strlen(ename));
P9_DPRINTK(P9_DEBUG_9P, "<<< RERROR (%d) %s\n", -ecode, P9_DPRINTK(P9_DEBUG_9P, "<<< RERROR (%d) %s\n",
ename); -ecode, ename);
kfree(ename);
} }
kfree(ename);
} else { } else {
err = p9pdu_readf(req->rc, c->proto_version, "d", &ecode); err = p9pdu_readf(req->rc, c->proto_version, "d", &ecode);
err = -ecode; err = -ecode;
...@@ -527,7 +497,6 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) ...@@ -527,7 +497,6 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req)
P9_DPRINTK(P9_DEBUG_9P, "<<< RLERROR (%d)\n", -ecode); P9_DPRINTK(P9_DEBUG_9P, "<<< RLERROR (%d)\n", -ecode);
} }
return err; return err;
out_err: out_err:
...@@ -536,6 +505,110 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) ...@@ -536,6 +505,110 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req)
return err; return err;
} }
/**
* p9_check_zc_errors - check 9p packet for error return and process it
* @c: current client instance
* @req: request to parse and check for error conditions
* @in_hdrlen: Size of response protocol buffer.
*
* returns error code if one is discovered, otherwise returns 0
*
* this will have to be more complicated if we have multiple
* error packet types
*/
static int p9_check_zc_errors(struct p9_client *c, struct p9_req_t *req,
char *uidata, int in_hdrlen, int kern_buf)
{
int err;
int ecode;
int8_t type;
char *ename = NULL;
err = p9_parse_header(req->rc, NULL, &type, NULL, 0);
if (err) {
P9_DPRINTK(P9_DEBUG_ERROR, "couldn't parse header %d\n", err);
return err;
}
if (type != P9_RERROR && type != P9_RLERROR)
return 0;
if (!p9_is_proto_dotl(c)) {
/* Error is reported in string format */
uint16_t len;
/* 7 = header size for RERROR, 2 is the size of string len; */
int inline_len = in_hdrlen - (7 + 2);
/* Read the size of error string */
err = p9pdu_readf(req->rc, c->proto_version, "w", &len);
if (err)
goto out_err;
ename = kmalloc(len + 1, GFP_NOFS);
if (!ename) {
err = -ENOMEM;
goto out_err;
}
if (len <= inline_len) {
/* We have error in protocol buffer itself */
if (pdu_read(req->rc, ename, len)) {
err = -EFAULT;
goto out_free;
}
} else {
/*
* Part of the data is in user space buffer.
*/
if (pdu_read(req->rc, ename, inline_len)) {
err = -EFAULT;
goto out_free;
}
if (kern_buf) {
memcpy(ename + inline_len, uidata,
len - inline_len);
} else {
err = copy_from_user(ename + inline_len,
uidata, len - inline_len);
if (err) {
err = -EFAULT;
goto out_free;
}
}
}
ename[len] = 0;
if (p9_is_proto_dotu(c)) {
/* For dotu we also have error code */
err = p9pdu_readf(req->rc,
c->proto_version, "d", &ecode);
if (err)
goto out_free;
err = -ecode;
}
if (!err || !IS_ERR_VALUE(err)) {
err = p9_errstr2errno(ename, strlen(ename));
P9_DPRINTK(P9_DEBUG_9P, "<<< RERROR (%d) %s\n",
-ecode, ename);
}
kfree(ename);
} else {
err = p9pdu_readf(req->rc, c->proto_version, "d", &ecode);
err = -ecode;
P9_DPRINTK(P9_DEBUG_9P, "<<< RLERROR (%d)\n", -ecode);
}
return err;
out_free:
kfree(ename);
out_err:
P9_DPRINTK(P9_DEBUG_ERROR, "couldn't parse error%d\n", err);
return err;
}
static struct p9_req_t * static struct p9_req_t *
p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...); p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...);
...@@ -579,23 +652,12 @@ static int p9_client_flush(struct p9_client *c, struct p9_req_t *oldreq) ...@@ -579,23 +652,12 @@ static int p9_client_flush(struct p9_client *c, struct p9_req_t *oldreq)
return 0; return 0;
} }
/** static struct p9_req_t *p9_client_prepare_req(struct p9_client *c,
* p9_client_rpc - issue a request and wait for a response int8_t type, int req_size,
* @c: client session const char *fmt, va_list ap)
* @type: type of request
* @fmt: protocol format string (see protocol.c)
*
* Returns request structure (which client must free using p9_free_req)
*/
static struct p9_req_t *
p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
{ {
va_list ap;
int tag, err; int tag, err;
struct p9_req_t *req; struct p9_req_t *req;
unsigned long flags;
int sigpending;
P9_DPRINTK(P9_DEBUG_MUX, "client %p op %d\n", c, type); P9_DPRINTK(P9_DEBUG_MUX, "client %p op %d\n", c, type);
...@@ -607,12 +669,6 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...) ...@@ -607,12 +669,6 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
if ((c->status == BeginDisconnect) && (type != P9_TCLUNK)) if ((c->status == BeginDisconnect) && (type != P9_TCLUNK))
return ERR_PTR(-EIO); return ERR_PTR(-EIO);
if (signal_pending(current)) {
sigpending = 1;
clear_thread_flag(TIF_SIGPENDING);
} else
sigpending = 0;
tag = P9_NOTAG; tag = P9_NOTAG;
if (type != P9_TVERSION) { if (type != P9_TVERSION) {
tag = p9_idpool_get(c->tagpool); tag = p9_idpool_get(c->tagpool);
...@@ -620,18 +676,50 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...) ...@@ -620,18 +676,50 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
return ERR_PTR(-ENOMEM); return ERR_PTR(-ENOMEM);
} }
req = p9_tag_alloc(c, tag); req = p9_tag_alloc(c, tag, req_size);
if (IS_ERR(req)) if (IS_ERR(req))
return req; return req;
/* marshall the data */ /* marshall the data */
p9pdu_prepare(req->tc, tag, type); p9pdu_prepare(req->tc, tag, type);
va_start(ap, fmt);
err = p9pdu_vwritef(req->tc, c->proto_version, fmt, ap); err = p9pdu_vwritef(req->tc, c->proto_version, fmt, ap);
va_end(ap);
if (err) if (err)
goto reterr; goto reterr;
p9pdu_finalize(req->tc); p9pdu_finalize(req->tc);
return req;
reterr:
p9_free_req(c, req);
return ERR_PTR(err);
}
/**
* p9_client_rpc - issue a request and wait for a response
* @c: client session
* @type: type of request
* @fmt: protocol format string (see protocol.c)
*
* Returns request structure (which client must free using p9_free_req)
*/
static struct p9_req_t *
p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
{
va_list ap;
int sigpending, err;
unsigned long flags;
struct p9_req_t *req;
va_start(ap, fmt);
req = p9_client_prepare_req(c, type, c->msize, fmt, ap);
va_end(ap);
if (IS_ERR(req))
return req;
if (signal_pending(current)) {
sigpending = 1;
clear_thread_flag(TIF_SIGPENDING);
} else
sigpending = 0;
err = c->trans_mod->request(c, req); err = c->trans_mod->request(c, req);
if (err < 0) { if (err < 0) {
...@@ -639,18 +727,14 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...) ...@@ -639,18 +727,14 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
c->status = Disconnected; c->status = Disconnected;
goto reterr; goto reterr;
} }
/* Wait for the response */
P9_DPRINTK(P9_DEBUG_MUX, "wait %p tag: %d\n", req->wq, tag);
err = wait_event_interruptible(*req->wq, err = wait_event_interruptible(*req->wq,
req->status >= REQ_STATUS_RCVD); req->status >= REQ_STATUS_RCVD);
P9_DPRINTK(P9_DEBUG_MUX, "wait %p tag: %d returned %d\n",
req->wq, tag, err);
if (req->status == REQ_STATUS_ERROR) { if (req->status == REQ_STATUS_ERROR) {
P9_DPRINTK(P9_DEBUG_ERROR, "req_status error %d\n", req->t_err); P9_DPRINTK(P9_DEBUG_ERROR, "req_status error %d\n", req->t_err);
err = req->t_err; err = req->t_err;
} }
if ((err == -ERESTARTSYS) && (c->status == Connected)) { if ((err == -ERESTARTSYS) && (c->status == Connected)) {
P9_DPRINTK(P9_DEBUG_MUX, "flushing\n"); P9_DPRINTK(P9_DEBUG_MUX, "flushing\n");
sigpending = 1; sigpending = 1;
...@@ -663,13 +747,11 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...) ...@@ -663,13 +747,11 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
if (req->status == REQ_STATUS_RCVD) if (req->status == REQ_STATUS_RCVD)
err = 0; err = 0;
} }
if (sigpending) { if (sigpending) {
spin_lock_irqsave(&current->sighand->siglock, flags); spin_lock_irqsave(&current->sighand->siglock, flags);
recalc_sigpending(); recalc_sigpending();
spin_unlock_irqrestore(&current->sighand->siglock, flags); spin_unlock_irqrestore(&current->sighand->siglock, flags);
} }
if (err < 0) if (err < 0)
goto reterr; goto reterr;
...@@ -678,7 +760,92 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...) ...@@ -678,7 +760,92 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
P9_DPRINTK(P9_DEBUG_MUX, "exit: client %p op %d\n", c, type); P9_DPRINTK(P9_DEBUG_MUX, "exit: client %p op %d\n", c, type);
return req; return req;
} }
reterr:
P9_DPRINTK(P9_DEBUG_MUX,
"exit: client %p op %d error: %d\n", c, type, err);
p9_free_req(c, req);
return ERR_PTR(err);
}
/**
* p9_client_zc_rpc - issue a request and wait for a response
* @c: client session
* @type: type of request
* @uidata: user bffer that should be ued for zero copy read
* @uodata: user buffer that shoud be user for zero copy write
* @inlen: read buffer size
* @olen: write buffer size
* @hdrlen: reader header size, This is the size of response protocol data
* @fmt: protocol format string (see protocol.c)
*
* Returns request structure (which client must free using p9_free_req)
*/
static struct p9_req_t *p9_client_zc_rpc(struct p9_client *c, int8_t type,
char *uidata, char *uodata,
int inlen, int olen, int in_hdrlen,
int kern_buf, const char *fmt, ...)
{
va_list ap;
int sigpending, err;
unsigned long flags;
struct p9_req_t *req;
va_start(ap, fmt);
/*
* We allocate a inline protocol data of only 4k bytes.
* The actual content is passed in zero-copy fashion.
*/
req = p9_client_prepare_req(c, type, P9_ZC_HDR_SZ, fmt, ap);
va_end(ap);
if (IS_ERR(req))
return req;
if (signal_pending(current)) {
sigpending = 1;
clear_thread_flag(TIF_SIGPENDING);
} else
sigpending = 0;
/* If we are called with KERNEL_DS force kern_buf */
if (segment_eq(get_fs(), KERNEL_DS))
kern_buf = 1;
err = c->trans_mod->zc_request(c, req, uidata, uodata,
inlen, olen, in_hdrlen, kern_buf);
if (err < 0) {
if (err == -EIO)
c->status = Disconnected;
goto reterr;
}
if (req->status == REQ_STATUS_ERROR) {
P9_DPRINTK(P9_DEBUG_ERROR, "req_status error %d\n", req->t_err);
err = req->t_err;
}
if ((err == -ERESTARTSYS) && (c->status == Connected)) {
P9_DPRINTK(P9_DEBUG_MUX, "flushing\n");
sigpending = 1;
clear_thread_flag(TIF_SIGPENDING);
if (c->trans_mod->cancel(c, req))
p9_client_flush(c, req);
/* if we received the response anyway, don't signal error */
if (req->status == REQ_STATUS_RCVD)
err = 0;
}
if (sigpending) {
spin_lock_irqsave(&current->sighand->siglock, flags);
recalc_sigpending();
spin_unlock_irqrestore(&current->sighand->siglock, flags);
}
if (err < 0)
goto reterr;
err = p9_check_zc_errors(c, req, uidata, in_hdrlen, kern_buf);
if (!err) {
P9_DPRINTK(P9_DEBUG_MUX, "exit: client %p op %d\n", c, type);
return req;
}
reterr: reterr:
P9_DPRINTK(P9_DEBUG_MUX, "exit: client %p op %d error: %d\n", c, type, P9_DPRINTK(P9_DEBUG_MUX, "exit: client %p op %d error: %d\n", c, type,
err); err);
...@@ -1330,13 +1497,15 @@ int ...@@ -1330,13 +1497,15 @@ int
p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset, p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset,
u32 count) u32 count)
{ {
int err, rsize;
struct p9_client *clnt;
struct p9_req_t *req;
char *dataptr; char *dataptr;
int kernel_buf = 0;
struct p9_req_t *req;
struct p9_client *clnt;
int err, rsize, non_zc = 0;
P9_DPRINTK(P9_DEBUG_9P, ">>> TREAD fid %d offset %llu %d\n", fid->fid, P9_DPRINTK(P9_DEBUG_9P, ">>> TREAD fid %d offset %llu %d\n",
(long long unsigned) offset, count); fid->fid, (long long unsigned) offset, count);
err = 0; err = 0;
clnt = fid->clnt; clnt = fid->clnt;
...@@ -1348,13 +1517,24 @@ p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset, ...@@ -1348,13 +1517,24 @@ p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset,
rsize = count; rsize = count;
/* Don't bother zerocopy for small IO (< 1024) */ /* Don't bother zerocopy for small IO (< 1024) */
if (((clnt->trans_mod->pref & P9_TRANS_PREF_PAYLOAD_MASK) == if (clnt->trans_mod->zc_request && rsize > 1024) {
P9_TRANS_PREF_PAYLOAD_SEP) && (rsize > 1024)) { char *indata;
req = p9_client_rpc(clnt, P9_TREAD, "dqE", fid->fid, offset, if (data) {
rsize, data, udata); kernel_buf = 1;
indata = data;
} else
indata = (char *)udata;
/*
* response header len is 11
* PDU Header(7) + IO Size (4)
*/
req = p9_client_zc_rpc(clnt, P9_TREAD, indata, NULL, rsize, 0,
11, kernel_buf, "dqd", fid->fid,
offset, rsize);
} else { } else {
non_zc = 1;
req = p9_client_rpc(clnt, P9_TREAD, "dqd", fid->fid, offset, req = p9_client_rpc(clnt, P9_TREAD, "dqd", fid->fid, offset,
rsize); rsize);
} }
if (IS_ERR(req)) { if (IS_ERR(req)) {
err = PTR_ERR(req); err = PTR_ERR(req);
...@@ -1370,7 +1550,7 @@ p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset, ...@@ -1370,7 +1550,7 @@ p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset,
P9_DPRINTK(P9_DEBUG_9P, "<<< RREAD count %d\n", count); P9_DPRINTK(P9_DEBUG_9P, "<<< RREAD count %d\n", count);
P9_DUMP_PKT(1, req->rc); P9_DUMP_PKT(1, req->rc);
if (!req->tc->pbuf_size) { if (non_zc) {
if (data) { if (data) {
memmove(data, dataptr, count); memmove(data, dataptr, count);
} else { } else {
...@@ -1396,6 +1576,7 @@ p9_client_write(struct p9_fid *fid, char *data, const char __user *udata, ...@@ -1396,6 +1576,7 @@ p9_client_write(struct p9_fid *fid, char *data, const char __user *udata,
u64 offset, u32 count) u64 offset, u32 count)
{ {
int err, rsize; int err, rsize;
int kernel_buf = 0;
struct p9_client *clnt; struct p9_client *clnt;
struct p9_req_t *req; struct p9_req_t *req;
...@@ -1411,19 +1592,24 @@ p9_client_write(struct p9_fid *fid, char *data, const char __user *udata, ...@@ -1411,19 +1592,24 @@ p9_client_write(struct p9_fid *fid, char *data, const char __user *udata,
if (count < rsize) if (count < rsize)
rsize = count; rsize = count;
/* Don't bother zerocopy form small IO (< 1024) */ /* Don't bother zerocopy for small IO (< 1024) */
if (((clnt->trans_mod->pref & P9_TRANS_PREF_PAYLOAD_MASK) == if (clnt->trans_mod->zc_request && rsize > 1024) {
P9_TRANS_PREF_PAYLOAD_SEP) && (rsize > 1024)) { char *odata;
req = p9_client_rpc(clnt, P9_TWRITE, "dqE", fid->fid, offset, if (data) {
rsize, data, udata); kernel_buf = 1;
odata = data;
} else
odata = (char *)udata;
req = p9_client_zc_rpc(clnt, P9_TWRITE, NULL, odata, 0, rsize,
P9_ZC_HDR_SZ, kernel_buf, "dqd",
fid->fid, offset, rsize);
} else { } else {
if (data) if (data)
req = p9_client_rpc(clnt, P9_TWRITE, "dqD", fid->fid, req = p9_client_rpc(clnt, P9_TWRITE, "dqD", fid->fid,
offset, rsize, data); offset, rsize, data);
else else
req = p9_client_rpc(clnt, P9_TWRITE, "dqU", fid->fid, req = p9_client_rpc(clnt, P9_TWRITE, "dqU", fid->fid,
offset, rsize, udata); offset, rsize, udata);
} }
if (IS_ERR(req)) { if (IS_ERR(req)) {
err = PTR_ERR(req); err = PTR_ERR(req);
...@@ -1824,7 +2010,7 @@ EXPORT_SYMBOL_GPL(p9_client_xattrcreate); ...@@ -1824,7 +2010,7 @@ EXPORT_SYMBOL_GPL(p9_client_xattrcreate);
int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset) int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset)
{ {
int err, rsize; int err, rsize, non_zc = 0;
struct p9_client *clnt; struct p9_client *clnt;
struct p9_req_t *req; struct p9_req_t *req;
char *dataptr; char *dataptr;
...@@ -1842,13 +2028,18 @@ int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset) ...@@ -1842,13 +2028,18 @@ int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset)
if (count < rsize) if (count < rsize)
rsize = count; rsize = count;
if ((clnt->trans_mod->pref & P9_TRANS_PREF_PAYLOAD_MASK) == /* Don't bother zerocopy for small IO (< 1024) */
P9_TRANS_PREF_PAYLOAD_SEP) { if (clnt->trans_mod->zc_request && rsize > 1024) {
req = p9_client_rpc(clnt, P9_TREADDIR, "dqF", fid->fid, /*
offset, rsize, data); * response header len is 11
* PDU Header(7) + IO Size (4)
*/
req = p9_client_zc_rpc(clnt, P9_TREADDIR, data, NULL, rsize, 0,
11, 1, "dqd", fid->fid, offset, rsize);
} else { } else {
non_zc = 1;
req = p9_client_rpc(clnt, P9_TREADDIR, "dqd", fid->fid, req = p9_client_rpc(clnt, P9_TREADDIR, "dqd", fid->fid,
offset, rsize); offset, rsize);
} }
if (IS_ERR(req)) { if (IS_ERR(req)) {
err = PTR_ERR(req); err = PTR_ERR(req);
...@@ -1863,7 +2054,7 @@ int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset) ...@@ -1863,7 +2054,7 @@ int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset)
P9_DPRINTK(P9_DEBUG_9P, "<<< RREADDIR count %d\n", count); P9_DPRINTK(P9_DEBUG_9P, "<<< RREADDIR count %d\n", count);
if (!req->tc->pbuf_size && data) if (non_zc)
memmove(data, dataptr, count); memmove(data, dataptr, count);
p9_free_req(clnt, req); p9_free_req(clnt, req);
......
...@@ -81,7 +81,7 @@ void p9stat_free(struct p9_wstat *stbuf) ...@@ -81,7 +81,7 @@ void p9stat_free(struct p9_wstat *stbuf)
} }
EXPORT_SYMBOL(p9stat_free); EXPORT_SYMBOL(p9stat_free);
static size_t pdu_read(struct p9_fcall *pdu, void *data, size_t size) size_t pdu_read(struct p9_fcall *pdu, void *data, size_t size)
{ {
size_t len = min(pdu->size - pdu->offset, size); size_t len = min(pdu->size - pdu->offset, size);
memcpy(data, &pdu->sdata[pdu->offset], len); memcpy(data, &pdu->sdata[pdu->offset], len);
...@@ -108,26 +108,6 @@ pdu_write_u(struct p9_fcall *pdu, const char __user *udata, size_t size) ...@@ -108,26 +108,6 @@ pdu_write_u(struct p9_fcall *pdu, const char __user *udata, size_t size)
return size - len; return size - len;
} }
static size_t
pdu_write_urw(struct p9_fcall *pdu, const char *kdata, const char __user *udata,
size_t size)
{
BUG_ON(pdu->size > P9_IOHDRSZ);
pdu->pubuf = (char __user *)udata;
pdu->pkbuf = (char *)kdata;
pdu->pbuf_size = size;
return 0;
}
static size_t
pdu_write_readdir(struct p9_fcall *pdu, const char *kdata, size_t size)
{
BUG_ON(pdu->size > P9_READDIRHDRSZ);
pdu->pkbuf = (char *)kdata;
pdu->pbuf_size = size;
return 0;
}
/* /*
b - int8_t b - int8_t
w - int16_t w - int16_t
...@@ -459,26 +439,6 @@ p9pdu_vwritef(struct p9_fcall *pdu, int proto_version, const char *fmt, ...@@ -459,26 +439,6 @@ p9pdu_vwritef(struct p9_fcall *pdu, int proto_version, const char *fmt,
errcode = -EFAULT; errcode = -EFAULT;
} }
break; break;
case 'E':{
int32_t cnt = va_arg(ap, int32_t);
const char *k = va_arg(ap, const void *);
const char __user *u = va_arg(ap,
const void __user *);
errcode = p9pdu_writef(pdu, proto_version, "d",
cnt);
if (!errcode && pdu_write_urw(pdu, k, u, cnt))
errcode = -EFAULT;
}
break;
case 'F':{
int32_t cnt = va_arg(ap, int32_t);
const char *k = va_arg(ap, const void *);
errcode = p9pdu_writef(pdu, proto_version, "d",
cnt);
if (!errcode && pdu_write_readdir(pdu, k, cnt))
errcode = -EFAULT;
}
break;
case 'U':{ case 'U':{
int32_t count = va_arg(ap, int32_t); int32_t count = va_arg(ap, int32_t);
const char __user *udata = const char __user *udata =
...@@ -637,10 +597,6 @@ void p9pdu_reset(struct p9_fcall *pdu) ...@@ -637,10 +597,6 @@ void p9pdu_reset(struct p9_fcall *pdu)
{ {
pdu->offset = 0; pdu->offset = 0;
pdu->size = 0; pdu->size = 0;
pdu->private = NULL;
pdu->pubuf = NULL;
pdu->pkbuf = NULL;
pdu->pbuf_size = 0;
} }
int p9dirent_read(char *buf, int len, struct p9_dirent *dirent, int p9dirent_read(char *buf, int len, struct p9_dirent *dirent,
......
...@@ -32,3 +32,4 @@ int p9pdu_prepare(struct p9_fcall *pdu, int16_t tag, int8_t type); ...@@ -32,3 +32,4 @@ int p9pdu_prepare(struct p9_fcall *pdu, int16_t tag, int8_t type);
int p9pdu_finalize(struct p9_fcall *pdu); int p9pdu_finalize(struct p9_fcall *pdu);
void p9pdu_dump(int, struct p9_fcall *); void p9pdu_dump(int, struct p9_fcall *);
void p9pdu_reset(struct p9_fcall *pdu); void p9pdu_reset(struct p9_fcall *pdu);
size_t pdu_read(struct p9_fcall *pdu, void *data, size_t size);
...@@ -21,30 +21,25 @@ ...@@ -21,30 +21,25 @@
/** /**
* p9_release_req_pages - Release pages after the transaction. * p9_release_req_pages - Release pages after the transaction.
* @*private: PDU's private page of struct trans_rpage_info
*/ */
void void p9_release_pages(struct page **pages, int nr_pages)
p9_release_req_pages(struct trans_rpage_info *rpinfo)
{ {
int i = 0; int i = 0;
while (pages[i] && nr_pages--) {
while (rpinfo->rp_data[i] && rpinfo->rp_nr_pages--) { put_page(pages[i]);
put_page(rpinfo->rp_data[i]);
i++; i++;
} }
} }
EXPORT_SYMBOL(p9_release_req_pages); EXPORT_SYMBOL(p9_release_pages);
/** /**
* p9_nr_pages - Return number of pages needed to accommodate the payload. * p9_nr_pages - Return number of pages needed to accommodate the payload.
*/ */
int int p9_nr_pages(char *data, int len)
p9_nr_pages(struct p9_req_t *req)
{ {
unsigned long start_page, end_page; unsigned long start_page, end_page;
start_page = (unsigned long)req->tc->pubuf >> PAGE_SHIFT; start_page = (unsigned long)data >> PAGE_SHIFT;
end_page = ((unsigned long)req->tc->pubuf + req->tc->pbuf_size + end_page = ((unsigned long)data + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
PAGE_SIZE - 1) >> PAGE_SHIFT;
return end_page - start_page; return end_page - start_page;
} }
EXPORT_SYMBOL(p9_nr_pages); EXPORT_SYMBOL(p9_nr_pages);
...@@ -58,35 +53,17 @@ EXPORT_SYMBOL(p9_nr_pages); ...@@ -58,35 +53,17 @@ EXPORT_SYMBOL(p9_nr_pages);
* @nr_pages: number of pages to accommodate the payload * @nr_pages: number of pages to accommodate the payload
* @rw: Indicates if the pages are for read or write. * @rw: Indicates if the pages are for read or write.
*/ */
int
p9_payload_gup(struct p9_req_t *req, size_t *pdata_off, int *pdata_len,
int nr_pages, u8 rw)
{
uint32_t first_page_bytes = 0;
int32_t pdata_mapped_pages;
struct trans_rpage_info *rpinfo;
*pdata_off = (__force size_t)req->tc->pubuf & (PAGE_SIZE-1);
if (*pdata_off) int p9_payload_gup(char *data, int *nr_pages, struct page **pages, int write)
first_page_bytes = min(((size_t)PAGE_SIZE - *pdata_off), {
req->tc->pbuf_size); int nr_mapped_pages;
rpinfo = req->tc->private; nr_mapped_pages = get_user_pages_fast((unsigned long)data,
pdata_mapped_pages = get_user_pages_fast((unsigned long)req->tc->pubuf, *nr_pages, write, pages);
nr_pages, rw, &rpinfo->rp_data[0]); if (nr_mapped_pages <= 0)
if (pdata_mapped_pages <= 0) return nr_mapped_pages;
return pdata_mapped_pages;
rpinfo->rp_nr_pages = pdata_mapped_pages; *nr_pages = nr_mapped_pages;
if (*pdata_off) {
*pdata_len = first_page_bytes;
*pdata_len += min((req->tc->pbuf_size - *pdata_len),
((size_t)pdata_mapped_pages - 1) << PAGE_SHIFT);
} else {
*pdata_len = min(req->tc->pbuf_size,
(size_t)pdata_mapped_pages << PAGE_SHIFT);
}
return 0; return 0;
} }
EXPORT_SYMBOL(p9_payload_gup); EXPORT_SYMBOL(p9_payload_gup);
...@@ -12,21 +12,6 @@ ...@@ -12,21 +12,6 @@
* *
*/ */
/* TRUE if it is user context */ void p9_release_pages(struct page **, int);
#define P9_IS_USER_CONTEXT (!segment_eq(get_fs(), KERNEL_DS)) int p9_payload_gup(char *, int *, struct page **, int);
int p9_nr_pages(char *, int);
/**
* struct trans_rpage_info - To store mapped page information in PDU.
* @rp_alloc:Set if this structure is allocd, not a reuse unused space in pdu.
* @rp_nr_pages: Number of mapped pages
* @rp_data: Array of page pointers
*/
struct trans_rpage_info {
u8 rp_alloc;
int rp_nr_pages;
struct page *rp_data[0];
};
void p9_release_req_pages(struct trans_rpage_info *);
int p9_payload_gup(struct p9_req_t *, size_t *, int *, int, u8);
int p9_nr_pages(struct p9_req_t *);
...@@ -150,12 +150,10 @@ static void req_done(struct virtqueue *vq) ...@@ -150,12 +150,10 @@ static void req_done(struct virtqueue *vq)
while (1) { while (1) {
spin_lock_irqsave(&chan->lock, flags); spin_lock_irqsave(&chan->lock, flags);
rc = virtqueue_get_buf(chan->vq, &len); rc = virtqueue_get_buf(chan->vq, &len);
if (rc == NULL) { if (rc == NULL) {
spin_unlock_irqrestore(&chan->lock, flags); spin_unlock_irqrestore(&chan->lock, flags);
break; break;
} }
chan->ring_bufs_avail = 1; chan->ring_bufs_avail = 1;
spin_unlock_irqrestore(&chan->lock, flags); spin_unlock_irqrestore(&chan->lock, flags);
/* Wakeup if anyone waiting for VirtIO ring space. */ /* Wakeup if anyone waiting for VirtIO ring space. */
...@@ -163,17 +161,6 @@ static void req_done(struct virtqueue *vq) ...@@ -163,17 +161,6 @@ static void req_done(struct virtqueue *vq)
P9_DPRINTK(P9_DEBUG_TRANS, ": rc %p\n", rc); P9_DPRINTK(P9_DEBUG_TRANS, ": rc %p\n", rc);
P9_DPRINTK(P9_DEBUG_TRANS, ": lookup tag %d\n", rc->tag); P9_DPRINTK(P9_DEBUG_TRANS, ": lookup tag %d\n", rc->tag);
req = p9_tag_lookup(chan->client, rc->tag); req = p9_tag_lookup(chan->client, rc->tag);
if (req->tc->private) {
struct trans_rpage_info *rp = req->tc->private;
int p = rp->rp_nr_pages;
/*Release pages */
p9_release_req_pages(rp);
atomic_sub(p, &vp_pinned);
wake_up(&vp_wq);
if (rp->rp_alloc)
kfree(rp);
req->tc->private = NULL;
}
req->status = REQ_STATUS_RCVD; req->status = REQ_STATUS_RCVD;
p9_client_cb(chan->client, req); p9_client_cb(chan->client, req);
} }
...@@ -193,9 +180,8 @@ static void req_done(struct virtqueue *vq) ...@@ -193,9 +180,8 @@ static void req_done(struct virtqueue *vq)
* *
*/ */
static int static int pack_sg_list(struct scatterlist *sg, int start,
pack_sg_list(struct scatterlist *sg, int start, int limit, char *data, int limit, char *data, int count)
int count)
{ {
int s; int s;
int index = start; int index = start;
...@@ -224,31 +210,36 @@ static int p9_virtio_cancel(struct p9_client *client, struct p9_req_t *req) ...@@ -224,31 +210,36 @@ static int p9_virtio_cancel(struct p9_client *client, struct p9_req_t *req)
* this takes a list of pages. * this takes a list of pages.
* @sg: scatter/gather list to pack into * @sg: scatter/gather list to pack into
* @start: which segment of the sg_list to start at * @start: which segment of the sg_list to start at
* @pdata_off: Offset into the first page
* @**pdata: a list of pages to add into sg. * @**pdata: a list of pages to add into sg.
* @nr_pages: number of pages to pack into the scatter/gather list
* @data: data to pack into scatter/gather list
* @count: amount of data to pack into the scatter/gather list * @count: amount of data to pack into the scatter/gather list
*/ */
static int static int
pack_sg_list_p(struct scatterlist *sg, int start, int limit, size_t pdata_off, pack_sg_list_p(struct scatterlist *sg, int start, int limit,
struct page **pdata, int count) struct page **pdata, int nr_pages, char *data, int count)
{ {
int s; int i = 0, s;
int i = 0; int data_off;
int index = start; int index = start;
if (pdata_off) { BUG_ON(nr_pages > (limit - start));
s = min((int)(PAGE_SIZE - pdata_off), count); /*
sg_set_page(&sg[index++], pdata[i++], s, pdata_off); * if the first page doesn't start at
count -= s; * page boundary find the offset
} */
data_off = offset_in_page(data);
while (count) { while (nr_pages) {
BUG_ON(index > limit); s = rest_of_page(data);
s = min((int)PAGE_SIZE, count); if (s > count)
sg_set_page(&sg[index++], pdata[i++], s, 0); s = count;
sg_set_page(&sg[index++], pdata[i++], s, data_off);
data_off = 0;
data += s;
count -= s; count -= s;
nr_pages--;
} }
return index-start; return index - start;
} }
/** /**
...@@ -261,114 +252,166 @@ pack_sg_list_p(struct scatterlist *sg, int start, int limit, size_t pdata_off, ...@@ -261,114 +252,166 @@ pack_sg_list_p(struct scatterlist *sg, int start, int limit, size_t pdata_off,
static int static int
p9_virtio_request(struct p9_client *client, struct p9_req_t *req) p9_virtio_request(struct p9_client *client, struct p9_req_t *req)
{ {
int in, out, inp, outp; int err;
struct virtio_chan *chan = client->trans; int in, out;
unsigned long flags; unsigned long flags;
size_t pdata_off = 0; struct virtio_chan *chan = client->trans;
struct trans_rpage_info *rpinfo = NULL;
int err, pdata_len = 0;
P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio request\n"); P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio request\n");
req->status = REQ_STATUS_SENT; req->status = REQ_STATUS_SENT;
req_retry:
spin_lock_irqsave(&chan->lock, flags);
/* Handle out VirtIO ring buffers */
out = pack_sg_list(chan->sg, 0,
VIRTQUEUE_NUM, req->tc->sdata, req->tc->size);
if (req->tc->pbuf_size && (req->tc->pubuf && P9_IS_USER_CONTEXT)) { in = pack_sg_list(chan->sg, out,
int nr_pages = p9_nr_pages(req); VIRTQUEUE_NUM, req->rc->sdata, req->rc->capacity);
int rpinfo_size = sizeof(struct trans_rpage_info) +
sizeof(struct page *) * nr_pages;
if (atomic_read(&vp_pinned) >= chan->p9_max_pages) { err = virtqueue_add_buf(chan->vq, chan->sg, out, in, req->tc);
err = wait_event_interruptible(vp_wq, if (err < 0) {
atomic_read(&vp_pinned) < chan->p9_max_pages); if (err == -ENOSPC) {
chan->ring_bufs_avail = 0;
spin_unlock_irqrestore(&chan->lock, flags);
err = wait_event_interruptible(*chan->vc_wq,
chan->ring_bufs_avail);
if (err == -ERESTARTSYS) if (err == -ERESTARTSYS)
return err; return err;
P9_DPRINTK(P9_DEBUG_TRANS, "9p: May gup pages now.\n");
}
if (rpinfo_size <= (req->tc->capacity - req->tc->size)) { P9_DPRINTK(P9_DEBUG_TRANS, "9p:Retry virtio request\n");
/* We can use sdata */ goto req_retry;
req->tc->private = req->tc->sdata + req->tc->size;
rpinfo = (struct trans_rpage_info *)req->tc->private;
rpinfo->rp_alloc = 0;
} else { } else {
req->tc->private = kmalloc(rpinfo_size, GFP_NOFS); spin_unlock_irqrestore(&chan->lock, flags);
if (!req->tc->private) { P9_DPRINTK(P9_DEBUG_TRANS,
P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: " "9p debug: "
"private kmalloc returned NULL"); "virtio rpc add_buf returned failure");
return -ENOMEM; return -EIO;
}
rpinfo = (struct trans_rpage_info *)req->tc->private;
rpinfo->rp_alloc = 1;
} }
}
virtqueue_kick(chan->vq);
spin_unlock_irqrestore(&chan->lock, flags);
err = p9_payload_gup(req, &pdata_off, &pdata_len, nr_pages, P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio request kicked\n");
req->tc->id == P9_TREAD ? 1 : 0); return 0;
if (err < 0) { }
if (rpinfo->rp_alloc)
kfree(rpinfo); static int p9_get_mapped_pages(struct virtio_chan *chan,
struct page **pages, char *data,
int nr_pages, int write, int kern_buf)
{
int err;
if (!kern_buf) {
/*
* We allow only p9_max_pages pinned. We wait for the
* Other zc request to finish here
*/
if (atomic_read(&vp_pinned) >= chan->p9_max_pages) {
err = wait_event_interruptible(vp_wq,
(atomic_read(&vp_pinned) < chan->p9_max_pages));
if (err == -ERESTARTSYS)
return err;
}
err = p9_payload_gup(data, &nr_pages, pages, write);
if (err < 0)
return err; return err;
} else { atomic_add(nr_pages, &vp_pinned);
atomic_add(rpinfo->rp_nr_pages, &vp_pinned); } else {
/* kernel buffer, no need to pin pages */
int s, index = 0;
int count = nr_pages;
while (nr_pages) {
s = rest_of_page(data);
pages[index++] = virt_to_page(data);
data += s;
nr_pages--;
} }
nr_pages = count;
} }
return nr_pages;
}
req_retry_pinned: /**
spin_lock_irqsave(&chan->lock, flags); * p9_virtio_zc_request - issue a zero copy request
* @client: client instance issuing the request
* @req: request to be issued
* @uidata: user bffer that should be ued for zero copy read
* @uodata: user buffer that shoud be user for zero copy write
* @inlen: read buffer size
* @olen: write buffer size
* @hdrlen: reader header size, This is the size of response protocol data
*
*/
static int
p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req,
char *uidata, char *uodata, int inlen,
int outlen, int in_hdr_len, int kern_buf)
{
int in, out, err;
unsigned long flags;
int in_nr_pages = 0, out_nr_pages = 0;
struct page **in_pages = NULL, **out_pages = NULL;
struct virtio_chan *chan = client->trans;
/* Handle out VirtIO ring buffers */ P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio request\n");
out = pack_sg_list(chan->sg, 0, VIRTQUEUE_NUM, req->tc->sdata,
req->tc->size); if (uodata) {
out_nr_pages = p9_nr_pages(uodata, outlen);
if (req->tc->pbuf_size && (req->tc->id == P9_TWRITE)) { out_pages = kmalloc(sizeof(struct page *) * out_nr_pages,
/* We have additional write payload buffer to take care */ GFP_NOFS);
if (req->tc->pubuf && P9_IS_USER_CONTEXT) { if (!out_pages) {
outp = pack_sg_list_p(chan->sg, out, VIRTQUEUE_NUM, err = -ENOMEM;
pdata_off, rpinfo->rp_data, pdata_len); goto err_out;
} else { }
char *pbuf; out_nr_pages = p9_get_mapped_pages(chan, out_pages, uodata,
if (req->tc->pubuf) out_nr_pages, 0, kern_buf);
pbuf = (__force char *) req->tc->pubuf; if (out_nr_pages < 0) {
else err = out_nr_pages;
pbuf = req->tc->pkbuf; kfree(out_pages);
outp = pack_sg_list(chan->sg, out, VIRTQUEUE_NUM, pbuf, out_pages = NULL;
req->tc->pbuf_size); goto err_out;
} }
out += outp;
} }
if (uidata) {
/* Handle in VirtIO ring buffers */ in_nr_pages = p9_nr_pages(uidata, inlen);
if (req->tc->pbuf_size && in_pages = kmalloc(sizeof(struct page *) * in_nr_pages,
((req->tc->id == P9_TREAD) || (req->tc->id == P9_TREADDIR))) { GFP_NOFS);
/* if (!in_pages) {
* Take care of additional Read payload. err = -ENOMEM;
* 11 is the read/write header = PDU Header(7) + IO Size (4). goto err_out;
* Arrange in such a way that server places header in the }
* alloced memory and payload onto the user buffer. in_nr_pages = p9_get_mapped_pages(chan, in_pages, uidata,
*/ in_nr_pages, 1, kern_buf);
inp = pack_sg_list(chan->sg, out, if (in_nr_pages < 0) {
VIRTQUEUE_NUM, req->rc->sdata, 11); err = in_nr_pages;
/* kfree(in_pages);
* Running executables in the filesystem may result in in_pages = NULL;
* a read request with kernel buffer as opposed to user buffer. goto err_out;
*/
if (req->tc->pubuf && P9_IS_USER_CONTEXT) {
in = pack_sg_list_p(chan->sg, out+inp, VIRTQUEUE_NUM,
pdata_off, rpinfo->rp_data, pdata_len);
} else {
char *pbuf;
if (req->tc->pubuf)
pbuf = (__force char *) req->tc->pubuf;
else
pbuf = req->tc->pkbuf;
in = pack_sg_list(chan->sg, out+inp, VIRTQUEUE_NUM,
pbuf, req->tc->pbuf_size);
} }
in += inp;
} else {
in = pack_sg_list(chan->sg, out, VIRTQUEUE_NUM,
req->rc->sdata, req->rc->capacity);
} }
req->status = REQ_STATUS_SENT;
req_retry_pinned:
spin_lock_irqsave(&chan->lock, flags);
/* out data */
out = pack_sg_list(chan->sg, 0,
VIRTQUEUE_NUM, req->tc->sdata, req->tc->size);
if (out_pages)
out += pack_sg_list_p(chan->sg, out, VIRTQUEUE_NUM,
out_pages, out_nr_pages, uodata, outlen);
/*
* Take care of in data
* For example TREAD have 11.
* 11 is the read/write header = PDU Header(7) + IO Size (4).
* Arrange in such a way that server places header in the
* alloced memory and payload onto the user buffer.
*/
in = pack_sg_list(chan->sg, out,
VIRTQUEUE_NUM, req->rc->sdata, in_hdr_len);
if (in_pages)
in += pack_sg_list_p(chan->sg, out + in, VIRTQUEUE_NUM,
in_pages, in_nr_pages, uidata, inlen);
err = virtqueue_add_buf(chan->vq, chan->sg, out, in, req->tc); err = virtqueue_add_buf(chan->vq, chan->sg, out, in, req->tc);
if (err < 0) { if (err < 0) {
...@@ -376,28 +419,45 @@ p9_virtio_request(struct p9_client *client, struct p9_req_t *req) ...@@ -376,28 +419,45 @@ p9_virtio_request(struct p9_client *client, struct p9_req_t *req)
chan->ring_bufs_avail = 0; chan->ring_bufs_avail = 0;
spin_unlock_irqrestore(&chan->lock, flags); spin_unlock_irqrestore(&chan->lock, flags);
err = wait_event_interruptible(*chan->vc_wq, err = wait_event_interruptible(*chan->vc_wq,
chan->ring_bufs_avail); chan->ring_bufs_avail);
if (err == -ERESTARTSYS) if (err == -ERESTARTSYS)
return err; goto err_out;
P9_DPRINTK(P9_DEBUG_TRANS, "9p:Retry virtio request\n"); P9_DPRINTK(P9_DEBUG_TRANS, "9p:Retry virtio request\n");
goto req_retry_pinned; goto req_retry_pinned;
} else { } else {
spin_unlock_irqrestore(&chan->lock, flags); spin_unlock_irqrestore(&chan->lock, flags);
P9_DPRINTK(P9_DEBUG_TRANS, P9_DPRINTK(P9_DEBUG_TRANS,
"9p debug: " "9p debug: "
"virtio rpc add_buf returned failure"); "virtio rpc add_buf returned failure");
if (rpinfo && rpinfo->rp_alloc) err = -EIO;
kfree(rpinfo); goto err_out;
return -EIO;
} }
} }
virtqueue_kick(chan->vq); virtqueue_kick(chan->vq);
spin_unlock_irqrestore(&chan->lock, flags); spin_unlock_irqrestore(&chan->lock, flags);
P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio request kicked\n"); P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio request kicked\n");
return 0; err = wait_event_interruptible(*req->wq,
req->status >= REQ_STATUS_RCVD);
/*
* Non kernel buffers are pinned, unpin them
*/
err_out:
if (!kern_buf) {
if (in_pages) {
p9_release_pages(in_pages, in_nr_pages);
atomic_sub(in_nr_pages, &vp_pinned);
}
if (out_pages) {
p9_release_pages(out_pages, out_nr_pages);
atomic_sub(out_nr_pages, &vp_pinned);
}
/* wakeup anybody waiting for slots to pin pages */
wake_up(&vp_wq);
}
kfree(in_pages);
kfree(out_pages);
return err;
} }
static ssize_t p9_mount_tag_show(struct device *dev, static ssize_t p9_mount_tag_show(struct device *dev,
...@@ -591,8 +651,8 @@ static struct p9_trans_module p9_virtio_trans = { ...@@ -591,8 +651,8 @@ static struct p9_trans_module p9_virtio_trans = {
.create = p9_virtio_create, .create = p9_virtio_create,
.close = p9_virtio_close, .close = p9_virtio_close,
.request = p9_virtio_request, .request = p9_virtio_request,
.zc_request = p9_virtio_zc_request,
.cancel = p9_virtio_cancel, .cancel = p9_virtio_cancel,
/* /*
* We leave one entry for input and one entry for response * We leave one entry for input and one entry for response
* headers. We also skip one more entry to accomodate, address * headers. We also skip one more entry to accomodate, address
...@@ -600,7 +660,6 @@ static struct p9_trans_module p9_virtio_trans = { ...@@ -600,7 +660,6 @@ static struct p9_trans_module p9_virtio_trans = {
* page in zero copy. * page in zero copy.
*/ */
.maxsize = PAGE_SIZE * (VIRTQUEUE_NUM - 3), .maxsize = PAGE_SIZE * (VIRTQUEUE_NUM - 3),
.pref = P9_TRANS_PREF_PAYLOAD_SEP,
.def = 0, .def = 0,
.owner = THIS_MODULE, .owner = THIS_MODULE,
}; };
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment