bus/pci: harmonize and document rte_pci_read_config return value
PATCH 1/2 bus/pci: harmonize and document rte_pci_read_config return value
On Linux, rte_pci_read_config on success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.
At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
--- drivers/bus/pci/bsd/pci.c | 4 +++- drivers/bus/pci/rte_bus_pci.h | 2 ++ 2 files changed, 5 insertions(+), 1 deletion(-)
drivers/bus/pci/bsd/pci.c +3 -1
@@ -439,6 +439,8 @@ int rte_pci_read_config(const struct rte_pci_device *dev, {int fd =-1 ;int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len;struct pci_io pi = { .pi_sel = { .pc_domain = dev->addr.domain,
@@ -469,7 +471,7 @@ int rte_pci_read_config(const struct rte_pci_device *dev, }close (fd); -return 0 ; +return return_len; error:if (fd >=0 )
drivers/bus/pci/rte_bus_pci.h +2 -0
@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver *driver);* The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config (const struct rte_pci_device *device,void *buf,size_t len,off_t offset);
Signature · 1 line
2.18.0
PATCH 2/2 virtio: fix PCI config err handling
From: Brian Russell <[email protected]>
In virtio_read_caps, rte_pci_read_config returns the number of bytes read from PCI config or < 0 on error.
If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
---
Follow-up from:
http://mails.dpdk.org/archives/dev/2017-June/067278.html
https://patches.dpdk.org/patch/25056/
drivers/net/virtio/virtio_pci.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-)
drivers/net/virtio/virtio_pci.c +7 -5
@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
Signature · 1 line
2.18.0
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 6 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_caps,rte_pci_read_configreturns the number of bytes read from PCI config or < 0 on error.
If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
Is this a fix or an improvement?
Or did you see anything broken without this patch? If so, we may need a fixes line and Cc stable.
Quoted 24 lines
Signed-off-by Brian Russell <[email protected]>---
Follow-up from:
http://mails.dpdk.org/archives/dev/2017-June/067278.html
https://patches.dpdk.org/patch/25056/drivers/net/virtio/virtio_pci.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-)drivers/net/virtio/virtio_pci.c +3
-2@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);
drivers/net/virtio/virtio_pci.c virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) · line 579
Quoted 10 lines
drivers/net/virtio/virtio_pci.c +2
-3return -1 ; }while (pos) { ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG,
Why change the log level to DEBUG?
Thanks
Quoted 8 lines
drivers/net/virtio/virtio_pci.c +2
-0+"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }Signature · 1 line
2.18.0
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 9 lines · Tiwei Bie
Quoted 7 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_caps,rte_pci_read_configreturns the number of bytes
read from PCI config or < 0 on error.
If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
Quoted 3 lines
Is this a fix or an improvement?
Or did you see anything broken without this patch? If so, we may need a fixes line and Cc stable.
It is a fix, as it was creating problems in production due to the constant flux of errors in the logs.
But given patch 1/2 is effectively doing a small change in the BSD bus API, and it's a requirement for 2/2, I don't think we can include it in the stable releases unfortunately.
Quoted 44 lines
Quoted 39 lines
Signed-off-by Brian Russell <[email protected]>---
Follow-up from:
http://mails.dpdk.org/archives/dev/2017-June/067278.html
https://patches.dpdk.org/patch/25056/drivers/net/virtio/virtio_pci.c | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) }
ret = rte_pci_read_config(dev, &pos, 1,PCI_CAPABILITY_LIST);drivers/net/virtio/virtio_pci.c +0
-2-if (ret <0 ) { - PMD_INIT_LOG(DEBUG, "failed to read pci capabilitylist");
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=1 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci capability list,ret %d", ret);
return -1;
}
while (pos) {
ret =rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos:%x", pos);
drivers/net/virtio/virtio_pci.c virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) · line 579
Quoted 2 lines
drivers/net/virtio/virtio_pci.c +2
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG,Why change the log level to DEBUG?
Quoted 1 line
Thanks
Beforehand reading less than the required amount of bytes caused problems in the following code, so it warranted printing errors - but now it will not go ahead without the right amount of data, so it's not critical anymore to inform the user.
Main issue is, log will get very spammy with errors, and paying customers don't like that :-)
Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 10 lines · Luca Boccassi
Quoted 9 lines · Tiwei Bie
Quoted 7 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_caps,rte_pci_read_configreturns the number of bytes
read from PCI config or < 0 on error.
If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
Quoted 6 lines
Quoted 3 lines
Is this a fix or an improvement?
Or did you see anything broken without this patch? If so, we may need a fixes line and Cc stable.It is a fix, as it was creating problems in production due to the constant flux of errors in the logs.
Could you be a bit more specific about which errors were logged if possible?
If my understanding is correct, you mean the errors were logged because less than the required amount of bytes were read?
Quoted 3 lines
But given patch 1/2 is effectively doing a small change in the BSD bus API, and it's a requirement for 2/2, I don't think we can include it in the stable releases unfortunately.
If it's a fix, we need a fixes line.
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
[...]
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
(quoted diff) +0
-0@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) }
ret =rte_pci_read_config(dev, &pos, 1, PCI_CAPABILITY_LIST);drivers/net/virtio/virtio_pci.c +3
-2-if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , + "failed to read pci capability list, ret %d", ret);return -1;
Quoted 8 lines
Quoted 8 lines
drivers/net/virtio/virtio_pci.c virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) · line 578
Quoted 8 lines
}
while (pos) {
ret =rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +1
-3-if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) {
Above code has to successfully read a full virtio PCI capability during each read, otherwise it will give up reading other capabilities and may fallback to the legacy mode. In which case it will fail to read the requested amount of bytes? Should we try to read the generic PCI fields first?
Besides, you also need to update other calls to rte_pci_read_config(), e.g.:
https://github.com/DPDK/dpdk/blob/76b9d9de5c7d/drivers/net/virtio/virtio_pci.c#L696
Thanks
Quoted 16 lines
Quoted 5 lines
Review
Quoted 1 line
(quoted diff) +1
-0+ PMD_INIT_LOG(DEBUG,Why change the log level to DEBUG?
Thanks
Beforehand reading less than the required amount of bytes caused problems in the following code, so it warranted printing errors - but now it will not go ahead without the right amount of data, so it's not critical anymore to inform the user.
Main issue is, log will get very spammy with errors, and paying customers don't like that :-)Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 18 lines · Tiwei Bie
Quoted 10 lines · Luca Boccassi
Quoted 9 lines · Tiwei Bie
Quoted 7 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_caps,rte_pci_read_configreturns the number of bytes
read from PCI config or < 0 on error.
If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Quoted 6 lines
Quoted 3 lines
Is this a fix or an improvement?
Or did you see anything broken without this patch? If so, we may need a fixes line and Cc stable.It is a fix, as it was creating problems in production due to the constant flux of errors in the logs.
Quoted 6 lines
Could you be a bit more specific about which errors were logged if possible?
If my understanding is correct, you mean the errors were logged because less than the required amount of bytes were read?
Yes - rte_pci_read_config on Linux will return not just 0/-1, but the actual number of bytes read. If it's less than the required amount, the code then goes on and reads garbage, which causes errors later in the execution. Checking that we actually got the amount of data we need fixes this issue.
Quoted 7 lines
Quoted 5 lines
But given patch 1/2 is effectively doing a small change in the BSD bus
API, and it's a requirement for 2/2, I don't think we can include it in
the stable releases unfortunately.If it's a fix, we need a fixes line.
Sure, will send a v2.
Quoted 28 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
[...]
Quoted 24 lines
Quoted 24 lines
Quoted 24 lines
(quoted diff) +0
-0@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) }
ret = rte_pci_read_config(dev, &pos, 1,PCI_CAPABILITY_LIST);(quoted diff) +0
-2- if (ret < 0) { - PMD_INIT_LOG(DEBUG, "failed to read pcicapability list");
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=1 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci capabilitylist, ret %d", ret);
return -1;
}
while (pos) {
ret =rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap atpos: %x", pos);
Quoted 8 lines
Quoted 1 line
Quoted 1 line
drivers/net/virtio/virtio_pci.c virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) · line 578
Quoted 1 line
drivers/net/virtio/virtio_pci.c +1
-0+if (ret !=sizeof (cap)) {Above code has to successfully read a full virtio PCI capability during each read, otherwise it will give up reading other capabilities and may fallback to the legacy mode. In which case it will fail to read the requested amount of bytes? Should we try to read the generic PCI fields first?
I do not know what exactly causes less than required bytes to be read, but we have seen it happen in production (not 100% of the times though
- so I think it's worth keeping the structure as-is). As you said in that case it falls back to legacy mode which, in our experience in production deployments, then succeeds. That's why the error level print is undesired - because the code will actually work via the fallback, but the customers will see scary errors in the logs and open escalations :-)
Quoted 7 lines
Besides, you also need to update other calls to
rte_pci_read_config(), e.g.:https://github.com/DPDK/dpdk/blob/76b9d9de5c7d/drivers/net/virtio/vir
tio_pci.c#L696Thanks
Sure I will apply the same changes in v2.
Signature · 2 lines
Kind regards, Luca Boccassi
PATCH 1/2 bus/pci: harmonize and document rte_pci_read_config return value
In reply to Luca BoccassiOn Linux, rte_pci_read_config on success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.
At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
--- drivers/bus/pci/bsd/pci.c | 4 +++- drivers/bus/pci/rte_bus_pci.h | 2 ++ 2 files changed, 5 insertions(+), 1 deletion(-)
drivers/bus/pci/bsd/pci.c +3 -1
@@ -439,6 +439,8 @@ int rte_pci_read_config(const struct rte_pci_device *dev, {int fd =-1 ;int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len;struct pci_io pi = { .pi_sel = { .pc_domain = dev->addr.domain,
@@ -469,7 +471,7 @@ int rte_pci_read_config(const struct rte_pci_device *dev, }close (fd); -return 0 ; +return return_len; error:if (fd >=0 )
drivers/bus/pci/rte_bus_pci.h +2 -0
@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver *driver);* The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config (const struct rte_pci_device *device,void *buf,size_t len,off_t offset);
Signature · 1 line
2.18.0
PATCH 2/2 virtio: fix PCI config err handling
From: Brian Russell <[email protected]>
In virtio_read_caps and vtpci_msix_detect, rte_pci_read_config returns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
---
v2: handle additional rte_pci_read_config incomplete reads
drivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++------------ 1 file changed, 22 insertions(+), 13 deletions(-)
drivers/net/virtio/virtio_pci.c +22 -13
@@ -560,6 +560,7 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw)uint8_t pos;struct virtio_pci_cap cap;int ret; +uint32_t multiplier;if (rte_pci_map_device (dev)) {PMD_INIT_LOG (DEBUG ,"failed to map pci device!" );
@@ -567,16 +568,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg =get_cfg_addr (dev, &cap);break ;case VIRTIO_PCI_CAP_NOTIFY_CFG : -rte_pci_read_config (dev, &hw->notify_off_multiplier, -4 , pos +sizeof (cap)); - hw->notify_base =get_cfg_addr (dev, &cap); +/* Avoid half-reads into hw */ + ret =rte_pci_read_config (dev, &multiplier,4 , + pos +sizeof (cap)); +if (ret ==4 ) { + hw->notify_off_multiplier = multiplier; + hw->notify_base =get_cfg_addr (dev, &cap); + }break ;case VIRTIO_PCI_CAP_DEVICE_CFG : hw->dev_cfg =get_cfg_addr (dev, &cap);
@@ -693,16 +700,18 @@ vtpci_msix_detect(struct rte_pci_device *dev)int ret; ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return VIRTIO_MSIX_NONE ; }while (pos) { ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
Signature · 1 line
2.18.0
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 24 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++--------
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0
...
Quoted 19 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg = get_cfg_addr(dev, &cap); break;case
VIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +2
-2- 4, pos + sizeof(cap)); - hw->notify_base =get_cfg_addr (dev, &cap); +/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev, &multiplier,4,
drivers/net/virtio/virtio_pci.c +3
-0+ pos + sizeof(cap)); +if (ret ==4 ) { + hw->notify_off_multiplier =multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base = get_cfg_addr(dev,&cap);
drivers/net/virtio/virtio_pci.c +1
-0+ }
Quoted 3 lines
break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg=get_cfg_addr(dev, &cap);
Tiwei: not 100% sure what's the best way to handle a failure here, this will avoid writing to *hw in case of errors. Let me know if this is OK.
Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 43 lines · Luca Boccassi
Quoted 24 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++--------
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0...
Quoted 17 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg = get_cfg_addr(dev, &cap); break;case
VIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +2
-2- 4, pos + sizeof(cap)); - hw->notify_base =get_cfg_addr (dev, &cap); +/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev, &multiplier,4,
drivers/net/virtio/virtio_pci.c +3
-0+ pos + sizeof(cap)); +if (ret ==4 ) { + hw->notify_off_multiplier =multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base = get_cfg_addr(dev,
Quoted 8 lines
Quoted 5 lines
&cap);
drivers/net/virtio/virtio_pci.c +1
-0+ }break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg=get_cfg_addr(dev, &cap);Tiwei: not 100% sure what's the best way to handle a failure here, this will avoid writing to *hw in case of errors. Let me know if this is OK.
My concern is about reading the virtio capability directly. With this patch, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. I'm not quite sure whether it will bring any bad consequence. How about just reading the first two bytes first? Something like this:
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/pci.c#L1491-L1497
Thanks
PATCH v3 1/2 bus/pci: harmonize and document rte_pci_read_config return value
In reply to Luca BoccassiOn Linux, rte_pci_read_config on success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.
At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
--- drivers/bus/pci/bsd/pci.c | 4 +++- drivers/bus/pci/rte_bus_pci.h | 2 ++ 2 files changed, 5 insertions(+), 1 deletion(-)
drivers/bus/pci/bsd/pci.c +3 -1
@@ -439,6 +439,8 @@ int rte_pci_read_config(const struct rte_pci_device *dev, {int fd =-1 ;int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len;struct pci_io pi = { .pi_sel = { .pc_domain = dev->addr.domain,
@@ -469,7 +471,7 @@ int rte_pci_read_config(const struct rte_pci_device *dev, }close (fd); -return 0 ; +return return_len; error:if (fd >=0 )
drivers/bus/pci/rte_bus_pci.h +2 -0
@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver *driver);* The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config (const struct rte_pci_device *device,void *buf,size_t len,off_t offset);
Signature · 1 line
2.18.0
PATCH v3 2/2 virtio: fix PCI config err handling
From: Brian Russell <[email protected]>
In virtio_read_caps and vtpci_msix_detect, rte_pci_read_config returns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
---
v2: handle additional rte_pci_read_config incomplete reads
v3: do not handle rte_pci_read_config of virtio cap, added in v2,
as it's less clear what the right thing to do there is
drivers/net/virtio/virtio_pci.c | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-)
drivers/net/virtio/virtio_pci.c +14 -10
@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
@@ -693,16 +695,18 @@ vtpci_msix_detect(struct rte_pci_device *dev)int ret; ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return VIRTIO_MSIX_NONE ; }while (pos) { ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
Signature · 1 line
2.18.0
Re: [PATCH 2/2] virtio: fix PCI config err handling
In reply to Tiwei BieQuoted 71 lines · Tiwei Bie
Quoted 52 lines · Luca Boccassi
Quoted 25 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++----
----
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0...
Quoted 24 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg = get_cfg_addr(dev, &cap); break;case
VIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +0
-2- 4, pos + sizeof(cap)); - hw->notify_base = get_cfg_addr(dev,&cap);
drivers/net/virtio/virtio_pci.c +2
-0+/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev,&multiplier,
4,drivers/net/virtio/virtio_pci.c +3
-0+ pos + sizeof(cap)); +if (ret ==4 ) { + hw->notify_off_multiplier =multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base =get_cfg_addr(dev,
&cap);drivers/net/virtio/virtio_pci.c +1
-0+ }Quoted 8 lines
Quoted 3 lines
break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg=get_cfg_addr(dev, &cap);Tiwei: not 100% sure what's the best way to handle a failure here, this
will avoid writing to *hw in case of errors. Let me know if this is OK.My concern is about reading the virtio capability directly. With this patch, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. I'm not quite sure whether it will bring any bad consequence. How about just reading the first two bytes first? Something like this:
Quoted 2 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/pci.
c#L1491-L1497
I'm not sure, to be honest. That bit didn't give me trouble at all, so at this point I'd much rather leave it alone so that the maintainers can take care of it how they see fit, if necessary :-)
I've sent a v3 that removes that individual change.
Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 72 lines · Luca Boccassi
Quoted 71 lines · Tiwei Bie
Quoted 52 lines · Luca Boccassi
Quoted 25 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++----
----
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0...
Quoted 24 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg = get_cfg_addr(dev, &cap); break;case
VIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +0
-2- 4, pos + sizeof(cap)); - hw->notify_base = get_cfg_addr(dev,&cap);
drivers/net/virtio/virtio_pci.c +2
-0+/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev,&multiplier,
4,drivers/net/virtio/virtio_pci.c +3
-0+ pos + sizeof(cap)); +if (ret ==4 ) { + hw->notify_off_multiplier =multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base =get_cfg_addr(dev,
&cap);drivers/net/virtio/virtio_pci.c +1
-0+ }Quoted 8 lines
Quoted 3 lines
break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg=get_cfg_addr(dev, &cap);Tiwei: not 100% sure what's the best way to handle a failure here, this
will avoid writing to *hw in case of errors. Let me know if this is OK.My concern is about reading the virtio capability directly. With this patch, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. I'm not quite sure whether it will bring any bad consequence. How about just reading the first two bytes first? Something like this:
Quoted 8 lines
Quoted 2 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/pci.
c#L1491-L1497I'm not sure, to be honest. That bit didn't give me trouble at all, so at this point I'd much rather leave it alone so that the maintainers can take care of it how they see fit, if necessary :-)
I've sent a v3 that removes that individual change.
My concern isn't about the above change (which handled the errors when reading multiplier). In fact, above change looks good to me! :-) I mean the below changes in this patch:
while (pos) {
ret = rte_pci_read_config(dev, &cap, sizeof(cap), pos);
- if (ret < 0) {
- PMD_INIT_LOG(ERR,
- "failed to read pci cap at pos: %x", pos);
+ if (ret != sizeof(cap)) {
+ PMD_INIT_LOG(DEBUG,
+ "failed to read pci cap at pos: %x ret %d",
+ pos, ret);
break;
}
With this change, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. Maybe it would be better to read the first two bytes first and check whether it's the capability we want to parse (i.e. vendor capability and MSIX capability). Something like this:
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/pci.c#L1491-L1497
How do you think?
Thanks
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 121 lines · Tiwei Bie
Quoted 84 lines · Luca Boccassi
Quoted 79 lines · Tiwei Bie
Quoted 60 lines · Luca Boccassi
Quoted 28 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_config
returns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the
failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++
----
----
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0...
Quoted 29 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg = get_cfg_addr(dev, &cap); break;case
VIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +0
-1- 4, pos +sizeof(cap));
drivers/net/virtio/virtio_pci.c +0
-1- hw->notify_base = get_cfg_addr(dev,&cap);
drivers/net/virtio/virtio_pci.c +2
-0+/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev,&multiplier,
4,drivers/net/virtio/virtio_pci.c +3
-0+ pos + sizeof(cap)); +if (ret ==4 ) { + hw->notify_off_multiplier =multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base =get_cfg_addr(dev,
&cap);drivers/net/virtio/virtio_pci.c +1
-0+ }break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg= get_cfg_addr(dev,Quoted 8 lines
Quoted 1 line
&cap);
Tiwei: not 100% sure what's the best way to handle a failure here,
this
will avoid writing to *hw in case of errors. Let me know if this is
OK.My concern is about reading the virtio capability directly. With this patch, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. I'm not quite sure whether it will bring any bad consequence. How about just reading the first two bytes first? Something like this:
Quoted 3 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/
pci.
c#L1491-L1497Quoted 7 lines
I'm not sure, to be honest. That bit didn't give me trouble at all, so
at this point I'd much rather leave it alone so that the maintainers
can take care of it how they see fit, if necessary :-)I've sent a v3 that removes that individual change.
My concern isn't about the above change (which handled the errors when reading multiplier). In fact, above change looks good to me! :-) I mean the below changes in this patch:
while (pos) {
ret =rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos: %x",pos);
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap at pos:%x ret %d",
(quoted diff) +1
-0+ pos, ret);break;
}With this change, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. Maybe it would be better to read the first two bytes first and check whether it's the capability we want to parse (i.e. vendor capability and MSIX capability). Something like this:
Quoted 6 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/pci.
c#L1491-L1497How do you think?
Thanks
Ah I see what you mean now, sorry. Would something like the following be what you are looking for:
ret = rte_pci_read_config(dev, &cap, 2, pos);
if (ret != 2) {
PMD_INIT_LOG(DEBUG,
"failed to read pci cap at pos: %x ret %d",
pos, ret);
break;
}
if (cap.cap_vndr != PCI_CAP_ID_MSIX &&
cap.cap_vndr != PCI_CAP_ID_VNDR) {
goto next;
}
ret = rte_pci_read_config(dev, &cap, sizeof(cap), pos);
if (ret != sizeof(cap)) {
PMD_INIT_LOG(DEBUG,
"failed to read pci cap at pos: %x ret %d",
pos, ret);
break;
}Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH 2/2] virtio: fix PCI config err handling
Quoted 144 lines · Luca Boccassi
Quoted 121 lines · Tiwei Bie
Quoted 84 lines · Luca Boccassi
Quoted 79 lines · Tiwei Bie
Quoted 60 lines · Luca Boccassi
Quoted 28 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_config
returns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the
failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35 +++++++++++++++++++++
----
----
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0...
Quoted 29 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) hw->common_cfg = get_cfg_addr(dev, &cap); break;case
VIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +0
-1- 4, pos +sizeof(cap));
drivers/net/virtio/virtio_pci.c +0
-1- hw->notify_base = get_cfg_addr(dev,&cap);
drivers/net/virtio/virtio_pci.c +2
-0+/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev,&multiplier,
4,drivers/net/virtio/virtio_pci.c +3
-0+ pos + sizeof(cap)); +if (ret ==4 ) { + hw->notify_off_multiplier =multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base =get_cfg_addr(dev,
&cap);drivers/net/virtio/virtio_pci.c +1
-0+ }break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg= get_cfg_addr(dev,Quoted 8 lines
Quoted 1 line
&cap);
Tiwei: not 100% sure what's the best way to handle a failure here,
this
will avoid writing to *hw in case of errors. Let me know if this is
OK.My concern is about reading the virtio capability directly. With this patch, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. I'm not quite sure whether it will bring any bad consequence. How about just reading the first two bytes first? Something like this:
Quoted 3 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/
pci.
c#L1491-L1497Quoted 7 lines
I'm not sure, to be honest. That bit didn't give me trouble at all, so
at this point I'd much rather leave it alone so that the maintainers
can take care of it how they see fit, if necessary :-)I've sent a v3 that removes that individual change.
My concern isn't about the above change (which handled the errors when reading multiplier). In fact, above change looks good to me! :-) I mean the below changes in this patch:
while (pos) {
ret =rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos: %x",pos);
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap at pos:%x ret %d",
(quoted diff) +1
-0+ pos, ret);break;
}With this change, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. Maybe it would be better to read the first two bytes first and check whether it's the capability we want to parse (i.e. vendor capability and MSIX capability). Something like this:
Quoted 6 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/pci.
c#L1491-L1497How do you think?
Thanks
Ah I see what you mean now, sorry. Would something like the following be what you are looking for:
ret = rte_pci_read_config(dev, &cap, 2, pos); if (ret != 2) { PMD_INIT_LOG(DEBUG, "failed to read pci cap at pos: %x ret %d", pos, ret); break; } if (cap.cap_vndr != PCI_CAP_ID_MSIX && cap.cap_vndr != PCI_CAP_ID_VNDR) { goto next; }
Quoted 8 lines
ret = rte_pci_read_config(dev, &cap, sizeof(cap), pos);
if (ret != sizeof(cap)) {
PMD_INIT_LOG(DEBUG,
"failed to read pci cap at pos: %x ret %d",
pos, ret);
break;
}Yeah, that's what I want. :-)
Just one nit, we don't need to read it as a virtio cap when dealing with the MSIX cap. When dealing with the MSIX cap, we just need to read the next two bytes:
https://github.com/DPDK/dpdk/blob/7281cf520f89/drivers/net/virtio/virtio_pci.c#L584-L594
https://github.com/freebsd/freebsd/blob/59e4185a9358/sys/dev/pci/pci.c#L892
Thanks!
Quoted 3 lines
Signature · 2 lines
Kind regards, Luca Boccassi
PATCH v4 1/2 bus/pci: harmonize and document rte_pci_read_config return value
In reply to Luca BoccassiOn Linux, rte_pci_read_config on success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.
At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
--- drivers/bus/pci/bsd/pci.c | 4 +++- drivers/bus/pci/rte_bus_pci.h | 2 ++ 2 files changed, 5 insertions(+), 1 deletion(-)
drivers/bus/pci/bsd/pci.c +3 -1
@@ -439,6 +439,8 @@ int rte_pci_read_config(const struct rte_pci_device *dev, {int fd =-1 ;int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len;struct pci_io pi = { .pi_sel = { .pc_domain = dev->addr.domain,
@@ -469,7 +471,7 @@ int rte_pci_read_config(const struct rte_pci_device *dev, }close (fd); -return 0 ; +return return_len; error:if (fd >=0 )
drivers/bus/pci/rte_bus_pci.h +2 -0
@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver *driver);* The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config (const struct rte_pci_device *device,void *buf,size_t len,off_t offset);
Signature · 1 line
2.18.0
PATCH v4 2/2 virtio: fix PCI config err handling
From: Brian Russell <[email protected]>
In virtio_read_caps and vtpci_msix_detect, rte_pci_read_config returns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
---
v2: handle additional rte_pci_read_config incomplete reads
v3: do not handle rte_pci_read_config of virtio cap, added in v2,
as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and
skip the cap entirely if it's not what we are looking for.
drivers/net/virtio/virtio_pci.c | 57 ++++++++++++++++++++++++--------- 1 file changed, 42 insertions(+), 15 deletions(-)
drivers/net/virtio/virtio_pci.c +42 -15
@@ -567,16 +567,30 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { + ret =rte_pci_read_config (dev, &cap,2 , pos); +if (ret !=2 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret); +break ; + } +if (cap.cap_vndr !=PCI_CAP_ID_MSIX && + cap.cap_vndr !=PCI_CAP_ID_VNDR ) { + goto next; + } + ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
@@ -689,25 +703,38 @@ enum virtio_msix_status vtpci_msix_detect(struct rte_pci_device *dev) {uint8_t pos; -struct virtio_pci_cap cap;int ret; ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return VIRTIO_MSIX_NONE ; }while (pos) { - ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +uint8_t cap[2 ]; + + ret =rte_pci_read_config (dev, cap,sizeof (cap), pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; } -if (cap.cap_vndr ==PCI_CAP_ID_MSIX ) { -uint16_t flags = ((uint16_t *)&cap)[1 ]; +if (cap[0 ] ==PCI_CAP_ID_MSIX ) { +uint16_t flags; + + ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +sizeof (cap)); +if (ret !=sizeof (flags)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos:" +" %x ret %d" , pos +sizeof (cap), + ret); +break ; + }if (flags &PCI_MSIX_ENABLE )return VIRTIO_MSIX_ENABLED ;
@@ -715,7 +742,7 @@ vtpci_msix_detect(struct rte_pci_device *dev)return VIRTIO_MSIX_DISABLED ; } - pos = cap.cap_next; + pos = cap[1 ]; }return VIRTIO_MSIX_NONE ;
Signature · 1 line
2.18.0
Re: [PATCH 2/2] virtio: fix PCI config err handling
In reply to Tiwei BieQuoted 179 lines · Tiwei Bie
Quoted 162 lines · Luca Boccassi
Quoted 136 lines · Tiwei Bie
Quoted 95 lines · Luca Boccassi
Quoted 89 lines · Tiwei Bie
Quoted 70 lines · Luca Boccassi
Quoted 31 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_config
returns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log
the
failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete readsdrivers/net/virtio/virtio_pci.c | 35
+++++++++++++++++++++
----
----
----1 file changed, 22 insertions(+), 13 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0...
Quoted 31 lines
drivers/net/virtio/virtio_pci.c +0
-0@@ -610,9 +613,13 @@ virtio_read_caps(struct
rte_pci_device
*dev,
structvirtio_hw*hw)
hw->common_cfg=
get_cfg_addr(dev,
&cap);
break;
caseVIRTIO_PCI_CAP_NOTIFY_CFG:drivers/net/virtio/virtio_pci.c +0
-1- rte_pci_read_config(dev, &hw-Quoted 1 line
notify_off_multiplier,drivers/net/virtio/virtio_pci.c +0
-1- 4, pos +sizeof(cap));
drivers/net/virtio/virtio_pci.c +0
-1- hw->notify_base =get_cfg_addr(dev,
&cap);drivers/net/virtio/virtio_pci.c +2
-0+/* Avoid half-reads into hw */ + ret = rte_pci_read_config(dev,&multiplier,
4,drivers/net/virtio/virtio_pci.c +1
-0+ pos +sizeof(cap));
drivers/net/virtio/virtio_pci.c +2
-0+if (ret ==4 ) { + hw-Quoted 1 line
notify_off_multiplier=multiplier;
drivers/net/virtio/virtio_pci.c +1
-0+ hw->notify_base =get_cfg_addr(dev,
&cap);drivers/net/virtio/virtio_pci.c +1
-0+ }Quoted 4 lines
break;
caseVIRTIO_PCI_CAP_DEVICE_CFG:
hw->dev_cfg=get_cfg_addr(dev, &cap);Quoted 7 lines
Tiwei: not 100% sure what's the best way to handle a failure
here,
this
will avoid writing to *hw in case of errors. Let me know if this is
OK.My concern is about reading the virtio capability directly. With this patch, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. I'm not quite sure whether it will bring any bad consequence. How about just reading the first two bytes first? Something like this:
Quoted 4 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/
pci/
pci.
c#L1491-L1497Quoted 8 lines
I'm not sure, to be honest. That bit didn't give me trouble at all,
so
at this point I'd much rather leave it alone so that the maintainers
can take care of it how they see fit, if necessary :-)I've sent a v3 that removes that individual change.
My concern isn't about the above change (which handled the errors when reading multiplier). In fact, above change looks good to me! :-) I mean the below changes in this patch:
while (pos) {
ret = rte_pci_read_config(dev, &cap, sizeof(cap),
pos);(quoted diff) +0
-3- if (ret < 0) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos:%x",
pos);(quoted diff) +3
-0+ if (ret != sizeof(cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos:
%x ret %d",(quoted diff) +1
-0+ pos, ret);break;
}With this change, it will give up reading other capabilities when failed to read a full virtio PCI capability structure (the returned bytes are less than the expected bytes) even though when the capability it's reading isn't a virtio vendor specific capability. Maybe it would be better to read the first two bytes first and check whether it's the capability we want to parse (i.e. vendor capability and MSIX capability). Something like this:
Quoted 7 lines
https://github.com/freebsd/freebsd/blob/72445a41b3ff/sys/dev/pci/
pci.
c#L1491-L1497How do you think?
Thanks
Ah I see what you mean now, sorry. Would something like the following
be what you are looking for:ret = rte_pci_read_config(dev, &cap, 2, pos); if (ret != 2) { PMD_INIT_LOG(DEBUG, "failed to read pci cap at pos: %x ret %d", pos, ret); break; } if (cap.cap_vndr != PCI_CAP_ID_MSIX && cap.cap_vndr != PCI_CAP_ID_VNDR) { goto next; }Quoted 9 lines
ret = rte_pci_read_config(dev, &cap, sizeof(cap), pos); if (ret != sizeof(cap)) { PMD_INIT_LOG(DEBUG, "failed to read pci cap at pos: %x ret %d", pos, ret); break; }Yeah, that's what I want. :-)
Just one nit, we don't need to read it as a virtio cap when dealing with the MSIX cap. When dealing with the MSIX cap, we just need to read the next two bytes:
Quoted 6 lines
https://github.com/DPDK/dpdk/blob/7281cf520f89/drivers/net/virtio/vir
tio_pci.c#L584-L594
https://github.com/freebsd/freebsd/blob/59e4185a9358/sys/dev/pci/pci.
c#L892Thanks!
Ok, done, see v4.
Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH v4 2/2] virtio: fix PCI config err handling
In reply to Luca BoccassiQuoted 51 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2,
as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and
skip the cap entirely if it's not what we are looking for.drivers/net/virtio/virtio_pci.c | 57 ++++++++++++++++++++++++--------- 1 file changed, 42 insertions(+), 15 deletions(-)drivers/net/virtio/virtio_pci.c +15
-2@@ -567,16 +567,30 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { + ret =rte_pci_read_config (dev, &cap,2 , pos); +if (ret !=2 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret); +break ; + } +if (cap.cap_vndr !=PCI_CAP_ID_MSIX && + cap.cap_vndr !=PCI_CAP_ID_VNDR ) { + goto next; + } + ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos);
drivers/net/virtio/virtio_pci.c virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) · line 596
Quoted 10 lines
drivers/net/virtio/virtio_pci.c +4
-3-if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
It seems that I didn't make myself clear in my previous comments. I mean it's better to handle MSIX cap and virtio cap respectively in this function. Currently we're always reading them as virtio caps. As we are strictly requiring that _read_config() should return the required number of bytes, it's not perfect to require it to return "virtio cap size" of bytes while we're trying to read a MSIX cap. So please change the code to something similar to this:
while (pos) {
ret = rte_pci_read_config(dev, &cap, 2, pos);
if (ret != 2) {
PMD_INIT_LOG(DEBUG,
"failed to read pci cap at pos: %x ret %d",
pos, ret);
break;
}
if (cap.cap_vndr == PCI_CAP_ID_MSIX) {
/* Transitional devices would also have this capability,
* that's why we also check if msix is enabled.
* 1st byte is cap ID; 2nd byte is the position of next
* cap; next two bytes are the flags.
*/
uint16_t flags;
ret = rte_pci_read_config(dev, &flags, sizeof(flags),
pos + 2);
if (ret != sizeof(flags)) {
PMD_INIT_LOG(DEBUG,
"failed to read pci cap at pos: %x ret %d",
pos + 2, ret);
break;
}
if (flags & PCI_MSIX_ENABLE)
hw->use_msix = VIRTIO_MSIX_ENABLED;
else
hw->use_msix = VIRTIO_MSIX_DISABLED;
}
if (cap.cap_vndr != PCI_CAP_ID_VNDR) {
PMD_INIT_LOG(DEBUG,
"[%2x] skipping non VNDR cap id: %02x",
pos, cap.cap_vndr);
goto next;
}
ret = rte_pci_read_config(dev, &cap, sizeof(cap), pos);
if (ret != sizeof(cap)) {
PMD_INIT_LOG(DEBUG,
"failed to read pci cap at pos: %x ret %d",
pos, ret);
break;
}
PMD_INIT_LOG(DEBUG,
"[%2x] cfg type: %u, bar: %u, offset: %04x, len: %u",
pos, cap.cfg_type, cap.bar, cap.offset, cap.length);
switch (cap.cfg_type) {
......Quoted 32 lines
drivers/net/virtio/virtio_pci.c +10
-8@@ -689,25 +703,38 @@ enum virtio_msix_status vtpci_msix_detect(struct rte_pci_device *dev) {uint8_t pos; -struct virtio_pci_cap cap;int ret; ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return VIRTIO_MSIX_NONE ; }while (pos) { - ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +uint8_t cap[2 ]; + + ret =rte_pci_read_config (dev, cap,sizeof (cap), pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; } -if (cap.cap_vndr ==PCI_CAP_ID_MSIX ) {
drivers/net/virtio/virtio_pci.c enum virtio_msix_status · line 734
Quoted 10 lines
drivers/net/virtio/virtio_pci.c +9
-1-uint16_t flags = ((uint16_t *)&cap)[1 ]; +if (cap[0 ] ==PCI_CAP_ID_MSIX ) { +uint16_t flags; + + ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +sizeof (cap)); +if (ret !=sizeof (flags)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap at pos:" + " %x ret %d", pos + sizeof(cap),
There is a build error:
In file included from drivers/net/virtio/virtio_pci.c:15:drivers/net/virtio/virtio_pci.c: In function ‘vtpci_msix_detect’:drivers/net/virtio/virtio_logs.h:13:3: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 5 has type ‘long unsigned int’ [-Werror=format=]
"%s(): " fmt "\n", __func__, ##args)
^~~~~~~~drivers/net/virtio/virtio_pci.c:732:5: note: in expansion of macro ‘PMD_INIT_LOG’
PMD_INIT_LOG(DEBUG,
^~~~~~~~~~~~
drivers/net/virtio/virtio_pci.c:734:14: note: format string is defined here
" %x ret %d", pos + sizeof(cap),
~^
%lxQuoted 18 lines
drivers/net/virtio/virtio_pci.c +4
-1+ ret); +break ; + }if (flags &PCI_MSIX_ENABLE )return VIRTIO_MSIX_ENABLED ;@@ -715,7 +742,7 @@ vtpci_msix_detect(struct rte_pci_device *dev)return VIRTIO_MSIX_DISABLED ; } - pos = cap.cap_next; + pos = cap[1 ]; }return VIRTIO_MSIX_NONE ;Signature · 1 line
2.18.0
PATCH v5 1/2 bus/pci: harmonize and document rte_pci_read_config return value
In reply to Luca BoccassiOn Linux, rte_pci_read_config on success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.
At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
--- drivers/bus/pci/bsd/pci.c | 4 +++- drivers/bus/pci/rte_bus_pci.h | 2 ++ 2 files changed, 5 insertions(+), 1 deletion(-)
drivers/bus/pci/bsd/pci.c +3 -1
@@ -439,6 +439,8 @@ int rte_pci_read_config(const struct rte_pci_device *dev, {int fd =-1 ;int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len;struct pci_io pi = { .pi_sel = { .pc_domain = dev->addr.domain,
@@ -469,7 +471,7 @@ int rte_pci_read_config(const struct rte_pci_device *dev, }close (fd); -return 0 ; +return return_len; error:if (fd >=0 )
drivers/bus/pci/rte_bus_pci.h +2 -0
@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver *driver);* The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config (const struct rte_pci_device *device,void *buf,size_t len,off_t offset);
Signature · 1 line
2.18.0
PATCH v5 2/2 virtio: fix PCI config err handling
From: Brian Russell <[email protected]>
In virtio_read_caps and vtpci_msix_detect, rte_pci_read_config returns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
---
v2: handle additional rte_pci_read_config incomplete reads
v3: do not handle rte_pci_read_config of virtio cap, added in v2,
as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and
skip the cap entirely if it's not what we are looking for.
v5: fetch only 2 flags bytes if the vndr is PCI_CAP_ID_MSIX
drivers/net/virtio/virtio_pci.c | 66 ++++++++++++++++++++++++--------- 1 file changed, 49 insertions(+), 17 deletions(-)
drivers/net/virtio/virtio_pci.c +49 -17
@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { - ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); + ret =rte_pci_read_config (dev, &cap,2 , pos); +if (ret !=2 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
@@ -586,7 +588,16 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw)* 1st byte is cap ID; 2nd byte is the position of next * cap; next two bytes are the flags. */ -uint16_t flags = ((uint16_t *)&cap)[1 ]; +uint16_t flags; + + ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +2 ); +if (ret !=sizeof (flags)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos:" +" %x ret %d" , pos +2 , ret); +break ; + }if (flags &PCI_MSIX_ENABLE ) hw->use_msix =VIRTIO_MSIX_ENABLED ;
@@ -601,6 +612,14 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) goto next; } + ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret); +break ; + } +PMD_INIT_LOG (DEBUG ,"[%2x] cfg type: %u, bar: %u, offset: %04x, len: %u" , pos, cap.cfg_type, cap.bar, cap.offset, cap.length);
@@ -689,25 +708,38 @@ enum virtio_msix_status vtpci_msix_detect(struct rte_pci_device *dev) {uint8_t pos; -struct virtio_pci_cap cap;int ret; ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return VIRTIO_MSIX_NONE ; }while (pos) { - ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +uint8_t cap[2 ]; + + ret =rte_pci_read_config (dev, cap,sizeof (cap), pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; } -if (cap.cap_vndr ==PCI_CAP_ID_MSIX ) { -uint16_t flags = ((uint16_t *)&cap)[1 ]; +if (cap[0 ] ==PCI_CAP_ID_MSIX ) { +uint16_t flags; + + ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +sizeof (cap)); +if (ret !=sizeof (flags)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos:" +" %lx ret %d" , pos +sizeof (cap), + ret); +break ; + }if (flags &PCI_MSIX_ENABLE )return VIRTIO_MSIX_ENABLED ;
@@ -715,7 +747,7 @@ vtpci_msix_detect(struct rte_pci_device *dev)return VIRTIO_MSIX_DISABLED ; } - pos = cap.cap_next; + pos = cap[1 ]; }return VIRTIO_MSIX_NONE ;
Signature · 1 line
2.18.0
Re: [PATCH v4 2/2] virtio: fix PCI config err handling
In reply to Tiwei BieQuoted 74 lines · Tiwei Bie
Quoted 69 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2, as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and skip the cap entirely if it's not what we are looking for.drivers/net/virtio/virtio_pci.c | 57 ++++++++++++++++++++++++-----
----1 file changed, 42 insertions(+), 15 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0@@ -567,16 +567,30 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) }
ret = rte_pci_read_config(dev, &pos, 1,PCI_CAPABILITY_LIST);drivers/net/virtio/virtio_pci.c +0
-2-if (ret <0 ) { - PMD_INIT_LOG(DEBUG, "failed to read pci capabilitylist");
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=1 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci capability list,ret %d", ret);
return -1;
}
while (pos) {drivers/net/virtio/virtio_pci.c +4
-0+ ret =rte_pci_read_config (dev, &cap,2 , pos); +if (ret !=2 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos: %x ret %d",
drivers/net/virtio/virtio_pci.c +8
-0+ pos, ret); +break ; + } +if (cap.cap_vndr !=PCI_CAP_ID_MSIX && + cap.cap_vndr !=PCI_CAP_ID_VNDR ) { + goto next; + } +ret =
rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos:%x", pos);
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos: %x ret %d",
drivers/net/virtio/virtio_pci.c +1
-0+ pos, ret);Quoted 3 lines
break;
}
Quoted 8 lines
It seems that I didn't make myself clear in my previous comments. I mean it's better to handle MSIX cap and virtio cap respectively in this function. Currently we're always reading them as virtio caps. As we are strictly requiring that
_read_config()should return the required number of bytes, it's not perfect to require it to return "virtio cap size" of bytes while we're trying to read a MSIX cap. So please change the code to something similar to this:
Sorry, though you meant in the vtpci_msix_detect function, which I changed. Fixed in v5.
Quoted 55 lines
Quoted 51 lines
drivers/net/virtio/virtio_pci.c +0
-1@@ -689,25 +703,38 @@ enum virtio_msix_status vtpci_msix_detect(struct rte_pci_device *dev) { uint8_t pos; -struct virtio_pci_cap cap; int ret;
ret = rte_pci_read_config(dev, &pos, 1,PCI_CAPABILITY_LIST);drivers/net/virtio/virtio_pci.c +0
-2-if (ret <0 ) { - PMD_INIT_LOG(DEBUG, "failed to read pci capabilitylist");
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=1 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci capability list,ret %d", ret);
returnVIRTIO_MSIX_NONE;
}
while (pos) {drivers/net/virtio/virtio_pci.c +0
-1- ret = rte_pci_read_config(dev, &cap, sizeof(cap),pos);
drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos:%x", pos);
drivers/net/virtio/virtio_pci.c +3
-0+uint8_t cap[2 ]; + + ret = rte_pci_read_config(dev, cap, sizeof(cap),pos);
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos: %x ret %d",
drivers/net/virtio/virtio_pci.c +1
-0+ pos, ret);break;
}drivers/net/virtio/virtio_pci.c +4
-2-if (cap.cap_vndr ==PCI_CAP_ID_MSIX ) { -uint16_t flags = ((uint16_t *)&cap)[1 ]; +if (cap[0 ] ==PCI_CAP_ID_MSIX ) { +uint16_t flags; + + ret = rte_pci_read_config(dev, &flags,sizeof(flags),
drivers/net/virtio/virtio_pci.c +4
-0+ pos + sizeof(cap)); +if (ret !=sizeof (flags)) { + PMD_INIT_LOG(DEBUG, + "failed to read pcicap at pos:"
drivers/net/virtio/virtio_pci.c +1
-0+ " %x ret %d", pos +Quoted 1 line
sizeof(cap),
There is a build error:
Quoted 16 lines
In file included from
drivers/net/virtio/virtio_pci.c:15:drivers/net/virtio/virtio_pci.c: In function ‘vtpci_msix_detect’:drivers/net/virtio/virtio_logs.h:13:3: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 5 has type ‘long unsigned int’ [-Werror=format=]
"%s(): " fmt "\n",__func__, ##args)
^~~~~~~~drivers/net/virtio/virtio_pci.c:732:5: note: in expansion of macro ‘PMD_INIT_LOG’
PMD_INIT_LOG(DEBUG,
^~~~~~~~~~~~drivers/net/virtio/virtio_pci.c:734:14: note: format string is defined here
" %x ret %d", pos + sizeof(cap),
~^
%lx
Ok, changed to %lx - which GCC version was that with? Didn't get any warning with 6.3 on Debian 9.
Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH v5 2/2] virtio: fix PCI config err handling
In reply to Luca BoccassiI just noticed the title. It should be "net/virtio: xxx", instead of "virtio: xxx".
On Mon, Aug 27, 2018 at 05:52:40PM +0100, Luca Boccassi wrote:
[...]
drivers/net/virtio/virtio_pci.c enum virtio_msix_status · line 740
Quoted 7 lines
drivers/net/virtio/virtio_pci.c +7
-0+ ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +sizeof (cap)); +if (ret !=sizeof (flags)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos:" +" %lx ret %d" , pos +sizeof (cap), + ret);
In file included from drivers/net/virtio/virtio_pci.c:15:0:drivers/net/virtio/virtio_pci.c: In function ‘vtpci_msix_detect’:drivers/net/virtio/virtio_logs.h:13:3: error: format ‘%lx’ expects argument of type ‘long unsigned int’, but argument 5 has type ‘unsigned int’ [-Werror=format=]
"%s(): " fmt "\n", __func__, ##args)
^drivers/net/virtio/virtio_pci.c:737:5: note: in expansion of macro ‘PMD_INIT_LOG’
PMD_INIT_LOG(DEBUG,
^cc1: all warnings being treated as errors
I got above build issues in 32bit build.
Apart from that,
Thanks!
Re: [PATCH v4 2/2] virtio: fix PCI config err handling
In reply to Luca BoccassiQuoted 84 lines · Luca Boccassi
Quoted 74 lines · Tiwei Bie
Quoted 69 lines · Luca Boccassi
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2, as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and skip the cap entirely if it's not what we are looking for.drivers/net/virtio/virtio_pci.c | 57 ++++++++++++++++++++++++-----
----1 file changed, 42 insertions(+), 15 deletions(-)drivers/net/virtio/virtio_pci.c +0
-0@@ -567,16 +567,30 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) }
ret = rte_pci_read_config(dev, &pos, 1,PCI_CAPABILITY_LIST);drivers/net/virtio/virtio_pci.c +0
-2-if (ret <0 ) { - PMD_INIT_LOG(DEBUG, "failed to read pci capabilitylist");
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=1 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci capability list,ret %d", ret);
return -1;
}
while (pos) {drivers/net/virtio/virtio_pci.c +4
-0+ ret =rte_pci_read_config (dev, &cap,2 , pos); +if (ret !=2 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos: %x ret %d",
drivers/net/virtio/virtio_pci.c +8
-0+ pos, ret); +break ; + } +if (cap.cap_vndr !=PCI_CAP_ID_MSIX && + cap.cap_vndr !=PCI_CAP_ID_VNDR ) { + goto next; + } +ret =
rte_pci_read_config(dev, &cap, sizeof(cap), pos);drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos:%x", pos);
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos: %x ret %d",
drivers/net/virtio/virtio_pci.c +1
-0+ pos, ret);Quoted 3 lines
break;
}Quoted 8 lines
It seems that I didn't make myself clear in my previous comments. I mean it's better to handle MSIX cap and virtio cap respectively in this function. Currently we're always reading them as virtio caps. As we are strictly requiring that
_read_config()should return the required number of bytes, it's not perfect to require it to return "virtio cap size" of bytes while we're trying to read a MSIX cap. So please change the code to something similar to this:
Quoted 2 lines
Sorry, though you meant in the
vtpci_msix_detectfunction, which I changed. Fixed in v5.
Thanks!
Quoted 73 lines
Quoted 56 lines
Quoted 52 lines
drivers/net/virtio/virtio_pci.c +0
-1@@ -689,25 +703,38 @@ enum virtio_msix_status vtpci_msix_detect(struct rte_pci_device *dev) { uint8_t pos; -struct virtio_pci_cap cap; int ret;
ret = rte_pci_read_config(dev, &pos, 1,PCI_CAPABILITY_LIST);drivers/net/virtio/virtio_pci.c +0
-2-if (ret <0 ) { - PMD_INIT_LOG(DEBUG, "failed to read pci capabilitylist");
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=1 ) { + PMD_INIT_LOG(DEBUG, + "failed to read pci capability list,ret %d", ret);
returnVIRTIO_MSIX_NONE;
}
while (pos) {drivers/net/virtio/virtio_pci.c +0
-1- ret = rte_pci_read_config(dev, &cap, sizeof(cap),pos);
drivers/net/virtio/virtio_pci.c +0
-3-if (ret <0 ) { - PMD_INIT_LOG(ERR, - "failed to read pci cap at pos:%x", pos);
drivers/net/virtio/virtio_pci.c +3
-0+uint8_t cap[2 ]; + + ret = rte_pci_read_config(dev, cap, sizeof(cap),pos);
drivers/net/virtio/virtio_pci.c +3
-0+if (ret !=sizeof (cap)) { + PMD_INIT_LOG(DEBUG, + "failed to read pci cap atpos: %x ret %d",
drivers/net/virtio/virtio_pci.c +1
-0+ pos, ret);break;
}drivers/net/virtio/virtio_pci.c +4
-2-if (cap.cap_vndr ==PCI_CAP_ID_MSIX ) { -uint16_t flags = ((uint16_t *)&cap)[1 ]; +if (cap[0 ] ==PCI_CAP_ID_MSIX ) { +uint16_t flags; + + ret = rte_pci_read_config(dev, &flags,sizeof(flags),
drivers/net/virtio/virtio_pci.c +4
-0+ pos + sizeof(cap)); +if (ret !=sizeof (flags)) { + PMD_INIT_LOG(DEBUG, + "failed to read pcicap at pos:"
drivers/net/virtio/virtio_pci.c +1
-0+ " %x ret %d", pos +Quoted 1 line
sizeof(cap),
There is a build error:
Quoted 16 lines
In file included from
drivers/net/virtio/virtio_pci.c:15:drivers/net/virtio/virtio_pci.c: In function ‘vtpci_msix_detect’:drivers/net/virtio/virtio_logs.h:13:3: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 5 has type ‘long unsigned int’ [-Werror=format=]
"%s(): " fmt "\n",__func__, ##args)
^~~~~~~~drivers/net/virtio/virtio_pci.c:732:5: note: in expansion of macro ‘PMD_INIT_LOG’
PMD_INIT_LOG(DEBUG,
^~~~~~~~~~~~drivers/net/virtio/virtio_pci.c:734:14: note: format string is defined here
" %x ret %d", pos + sizeof(cap),
~^
%lx
Quoted 2 lines
Ok, changed to %lx - which GCC version was that with? Didn't get any warning with 6.3 on Debian 9.
I saw this with gcc (Debian 8.2.0-4) 8.2.0 on Debian sid. I also saw similar warnings with clang version 6.0.0 (tags/RELEASE_600/final 326565) on FreeBSD 11.2
Quoted 4 lines
Signature · 3 lines
-- Kind regards, Luca Boccassi
PATCH v6 1/2 bus/pci: harmonize and document rte_pci_read_config return value
In reply to Luca BoccassiOn Linux, rte_pci_read_config on success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.
At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
--- drivers/bus/pci/bsd/pci.c | 4 +++- drivers/bus/pci/rte_bus_pci.h | 2 ++ 2 files changed, 5 insertions(+), 1 deletion(-)
drivers/bus/pci/bsd/pci.c +3 -1
@@ -439,6 +439,8 @@ int rte_pci_read_config(const struct rte_pci_device *dev, {int fd =-1 ;int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len;struct pci_io pi = { .pi_sel = { .pc_domain = dev->addr.domain,
@@ -469,7 +471,7 @@ int rte_pci_read_config(const struct rte_pci_device *dev, }close (fd); -return 0 ; +return return_len; error:if (fd >=0 )
drivers/bus/pci/rte_bus_pci.h +2 -0
@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver *driver);* The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config (const struct rte_pci_device *device,void *buf,size_t len,off_t offset);
Signature · 1 line
2.18.0
PATCH v6 2/2 net/virtio: fix PCI config err handling
From: Brian Russell <[email protected]>
In virtio_read_caps and vtpci_msix_detect, rte_pci_read_config returns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.
---
v2: handle additional rte_pci_read_config incomplete reads
v3: do not handle rte_pci_read_config of virtio cap, added in v2,
as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and
skip the cap entirely if it's not what we are looking for.
v5: fetch only 2 flags bytes if the vndr is PCI_CAP_ID_MSIX
v6: fix 32bit build by changing the printf format specifier, fix patch title
drivers/net/virtio/virtio_pci.c | 65 ++++++++++++++++++++++++--------- 1 file changed, 48 insertions(+), 17 deletions(-)
drivers/net/virtio/virtio_pci.c +48 -17
@@ -567,16 +567,18 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) } ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return -1 ; }while (pos) { - ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); + ret =rte_pci_read_config (dev, &cap,2 , pos); +if (ret !=2 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; }
@@ -586,7 +588,16 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw)* 1st byte is cap ID; 2nd byte is the position of next * cap; next two bytes are the flags. */ -uint16_t flags = ((uint16_t *)&cap)[1 ]; +uint16_t flags; + + ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +2 ); +if (ret !=sizeof (flags)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos:" +" %x ret %d" , pos +2 , ret); +break ; + }if (flags &PCI_MSIX_ENABLE ) hw->use_msix =VIRTIO_MSIX_ENABLED ;
@@ -601,6 +612,14 @@ virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw) goto next; } + ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret); +break ; + } +PMD_INIT_LOG (DEBUG ,"[%2x] cfg type: %u, bar: %u, offset: %04x, len: %u" , pos, cap.cfg_type, cap.bar, cap.offset, cap.length);
@@ -689,25 +708,37 @@ enum virtio_msix_status vtpci_msix_detect(struct rte_pci_device *dev) {uint8_t pos; -struct virtio_pci_cap cap;int ret; ret =rte_pci_read_config (dev, &pos,1 ,PCI_CAPABILITY_LIST ); -if (ret <0 ) { -PMD_INIT_LOG (DEBUG ,"failed to read pci capability list" ); +if (ret !=1 ) { +PMD_INIT_LOG (DEBUG , +"failed to read pci capability list, ret %d" , ret);return VIRTIO_MSIX_NONE ; }while (pos) { - ret =rte_pci_read_config (dev, &cap,sizeof (cap), pos); -if (ret <0 ) { -PMD_INIT_LOG (ERR , -"failed to read pci cap at pos: %x" , pos); +uint8_t cap[2 ]; + + ret =rte_pci_read_config (dev, cap,sizeof (cap), pos); +if (ret !=sizeof (cap)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos: %x ret %d" , + pos, ret);break ; } -if (cap.cap_vndr ==PCI_CAP_ID_MSIX ) { -uint16_t flags = ((uint16_t *)&cap)[1 ]; +if (cap[0 ] ==PCI_CAP_ID_MSIX ) { +uint16_t flags; + + ret =rte_pci_read_config (dev, &flags,sizeof (flags), + pos +sizeof (cap)); +if (ret !=sizeof (flags)) { +PMD_INIT_LOG (DEBUG , +"failed to read pci cap at pos:" +" %x ret %d" , pos +2 , ret); +break ; + }if (flags &PCI_MSIX_ENABLE )return VIRTIO_MSIX_ENABLED ;
@@ -715,7 +746,7 @@ vtpci_msix_detect(struct rte_pci_device *dev)return VIRTIO_MSIX_DISABLED ; } - pos = cap.cap_next; + pos = cap[1 ]; }return VIRTIO_MSIX_NONE ;
Signature · 1 line
2.18.0
Re: [PATCH v5 2/2] virtio: fix PCI config err handling
In reply to Tiwei BieQuoted 2 lines · Tiwei Bie
I just noticed the title. It should be "net/virtio: xxx", instead of "virtio: xxx".
Fixed
Quoted 26 lines · Luca Boccassi
[...]
drivers/net/virtio/virtio_pci.c enum virtio_msix_status · line 740
Quoted 10 lines
(quoted diff) +1
-0+ ret = rte_pci_read_config(dev, &flags,sizeof(flags),
drivers/net/virtio/virtio_pci.c +4
-0+ pos + sizeof(cap)); +if (ret !=sizeof (flags)) { + PMD_INIT_LOG(DEBUG, + "failed to read pcicap at pos:"
(quoted diff) +1
-0+ " %lx ret %d", pos +sizeof(cap),
drivers/net/virtio/virtio_pci.c +1
-0+ ret);In file included from
drivers/net/virtio/virtio_pci.c:15:0:drivers/net/virtio/virtio_pci.c: In function ‘vtpci_msix_detect’:drivers/net/virtio/virtio_logs.h:13:3: error: format ‘%lx’ expects argument of type ‘long unsigned int’, but argument 5 has type ‘unsigned int’ [-Werror=format=]
"%s(): " fmt "\n",__func__, ##args)
^drivers/net/virtio/virtio_pci.c:737:5: note: in expansion of macro ‘PMD_INIT_LOG’
PMD_INIT_LOG(DEBUG,
^
cc1: all warnings being treated as errors
Quoted 8 lines
I got above build issues in 32bit build.
Apart from that,
Reviewed-by Tiwei Bie <[email protected]>Thanks!
So the 32 and 64 bit builds want opposite things, due to the sizeof type. I changed it to avoid sizeof and just calculate it like in every other prints of these functions to avoid the issue.
Signature · 2 lines
Kind regards, Luca Boccassi
Re: [PATCH v6 2/2] net/virtio: fix PCI config err handling
In reply to Luca Boccassi28/08/2018 12:12, Luca Boccassi:
Quoted 11 lines
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>
Quoted 8 lines
---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2,
as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and
skip the cap entirely if it's not what we are looking for.
v5: fetch only 2 flags bytes if the vndr isPCI_CAP_ID_MSIX
v6: fix 32bit build by changing the printf format specifier, fix patch title
Tiwei did a comment on v5 and provided his Reviewed-by. Is it OK to apply v6 with its tag?
All is fixed?
Re: [PATCH v6 2/2] net/virtio: fix PCI config err handling
Quoted 23 lines · Thomas Monjalon
28/08/2018 12:12, Luca Boccassi:
Quoted 12 lines
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>Quoted 9 lines
---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2, as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and skip the cap entirely if it's not what we are looking for.
v5: fetch only 2 flags bytes if the vndr isPCI_CAP_ID_MSIX
v6: fix 32bit build by changing the printf format specifier, fix patch title
Quoted 3 lines
Tiwei did a comment on v5 and provided his Reviewed-by. Is it OK to apply v6 with its tag?
All is fixed?
Hi,
The title has been fixed (virtio -> net/virtio) and I just tried a 32bit build and it works fine after the fix from v6.
Signature · 2 lines
Kind regards, Luca Boccassi
Tiwei Bie
Reviewed-byRe: [PATCH v6 2/2] net/virtio: fix PCI config err handling
Quoted 24 lines · Luca Boccassi
Quoted 23 lines · Thomas Monjalon
28/08/2018 12:12, Luca Boccassi:
Quoted 21 lines
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2, as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and skip the cap entirely if it's not what we are looking for.
v5: fetch only 2 flags bytes if the vndr isPCI_CAP_ID_MSIX
v6: fix 32bit build by changing the printf format specifier, fix patch title
Quoted 8 lines
Quoted 3 lines
Tiwei did a comment on v5 and provided his Reviewed-by. Is it OK to apply v6 with its tag?
All is fixed?Hi,
The title has been fixed (virtio -> net/virtio) and I just tried a 32bit build and it works fine after the fix from v6.
Thanks Luca!
Quoted 4 lines
Signature · 3 lines
-- Kind regards, Luca Boccassi
Quoted 51 lines · Luca Boccassi
On Linux,
rte_pci_read_configon success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
Signed-off-by Luca Boccassi <[email protected]>---
drivers/bus/pci/bsd/pci.c | 4 +++-
drivers/bus/pci/rte_bus_pci.h | 2 ++2 files changed, 5 insertions(+), 1 deletion(-)drivers/bus/pci/bsd/pci.c +2
-0@@ -439,6 +439,8 @@ int rte_pci_read_config(const structrte_pci_device *dev, { int fd = -1; int size; +/* Copy Linux implementation's behaviour */ +const int return_len = len; struct pci_io pi = { .pi_sel = {.pc_domain =
dev->addr.domain,drivers/bus/pci/bsd/pci.c +0
-0@@ -469,7 +471,7 @@ int rte_pci_read_config(const structrte_pci_device *dev, } close(fd);drivers/bus/pci/bsd/pci.c +1
-1-return 0 ; +return return_len;
error:
if (fd >= 0)drivers/bus/pci/rte_bus_pci.h +2
-0@@ -219,6 +219,8 @@ void rte_pci_unregister(struct rte_pci_driver*driver); * The length of the data buffer. * @param offset * The offset into PCI config space +* @return +* Number of bytes read on success, negative on error. */ int rte_pci_read_config(const struct rte_pci_device *device,
Quoted 1 line
void *buf,
size_tlen,off_toffset);
Hi Bruce,
Any chance you could please review the small BSD patch above when you have a sec? Thanks!
Signature · 2 lines
Kind regards, Luca Boccassi
Bruce Richardson
Acked-byQuoted 10 lines · Luca Boccassi
Quoted 9 lines · Luca Boccassi
On Linux,
rte_pci_read_configon success returns the number of read bytes, but on BSD it returns 0.
Document the return values, and have BSD behave as Linux does.At least one case (bnx2x PMD) treats 0 as an error, so the change makes sense also for that.
Signed-off-by Luca Boccassi <[email protected]>---
Re: [PATCH v6 2/2] net/virtio: fix PCI config err handling
In reply to Tiwei Bie11/10/2018 15:01, Tiwei Bie:
Quoted 29 lines · Luca Boccassi
Quoted 28 lines · Thomas Monjalon
Quoted 26 lines
28/08/2018 12:12, Luca Boccassi:
Quoted 12 lines
From: Brian Russell <[email protected]>
In
virtio_read_capsandvtpci_msix_detect,rte_pci_read_configreturns
the number of bytes read from PCI config or < 0 on error. If less than the expected number of bytes are read then log the failure and return rather than carrying on with garbage.Fixes 6ba1f63b5ab0 ("virtio: support specification 1.0")Signed-off-by Brian Russell <[email protected]>Signed-off-by Luca Boccassi <[email protected]>Quoted 9 lines
---
v2: handle additionalrte_pci_read_configincomplete reads
v3: do not handlerte_pci_read_configof virtio cap, added in v2,
as it's less clear what the right thing to do there is
v4: do a more robust check - first check what the vendor is, and
skip the cap entirely if it's not what we are looking for.
v5: fetch only 2 flags bytes if the vndr isPCI_CAP_ID_MSIX
v6: fix 32bit build by changing the printf format specifier, fix patch titleTiwei did a comment on v5 and provided his Reviewed-by. Is it OK to apply v6 with its tag?
All is fixed?
Quoted 8 lines
Quoted 4 lines
Hi,
The title has been fixed (virtio -> net/virtio) and I just tried a 32bit build and it works fine after the fix from v6.
Thanks Luca!
Reviewed-by Tiwei Bie <[email protected]>
Series applied, thanks