PCI: Safe VPD access
v4 cover PCI: Safe VPD access
PCI VPD suffers from two problems: it has a very rudimentary interface and it relies on correctly formatted data. And essentially it provides a direct channel into the card hardware. In other words, plenty of chances to mess things up.
With the original implementation we would just read the VPD space, blissfully ignorant of the data formatting of the VPD data. This would work if the VPD space happens to be properly implemented. However, I've had several reports where a simple 'cat' on the vpd attribute would return garbage (if you're lucky), trigger a timeout with a kernel warning (which actually triggered this patchset), or hang your machine (if you're really unlucky).
So this patchset validates the VPD data, setting the VPD attribute to the correct size or disabling VPD access altogether if no valid data is found.
Changes to v3:
- Do not calculate size for pci function devices
- Do not disable the attribute on error
- move size calculation into the pci_vpd_ops structure
Hannes Reinecke (2):
pci: Update VPD definitions
pci: Update VPD size with correct length
drivers/pci/access.c | 81 ++++++++++++++++++++++++++++++++++++++++++++++++++++ drivers/pci/pci.h | 1 + include/linux/pci.h | 27 ++++++++++++++++-- 3 files changed, 107 insertions(+), 2 deletions(-)
Signature · 1 line
1.8.5.6
PATCH v4 1/2 pci: Update VPD definitions
The 'end' tag is actually 0x0f, it's the representation as a small resource data type tag that's 0x78 (ie shifted by 3). This patch also adds helper functions to extract the resource data type tags for both large and small resource data types.
--- include/linux/pci.h | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)
include/linux/pci.h +25 -2
@@ -1827,12 +1827,13 @@ bool pci_acs_path_enabled(struct pci_dev *start,#define PCI_VPD_LRDT_RW_DATA PCI_VPD_LRDT_ID(PCI_VPD_LTIN_RW_DATA)/* Small Resource Data Type Tag Item Names */ -#define PCI_VPD_STIN_END 0x78/* End */ +#define PCI_VPD_STIN_END 0x0f/* End */ -#define PCI_VPD_SRDT_END PCI_VPD_STIN_END +#define PCI_VPD_SRDT_END (PCI_VPD_STIN_END << 3)#define PCI_VPD_SRDT_TIN_MASK 0x78#define PCI_VPD_SRDT_LEN_MASK 0x07 +#define PCI_VPD_LRDT_TIN_MASK 0x7f#define PCI_VPD_LRDT_TAG_SIZE 3#define PCI_VPD_SRDT_TAG_SIZE 1
@@ -1856,6 +1857,17 @@ static inline u16 pci_vpd_lrdt_size(const u8 *lrdt) }/** +* pci_vpd_lrdt_tag - Extracts the Large Resource Data Type Tag Item +* @lrdt: Pointer to the beginning of the Large Resource Data Type tag +* +* Returns the extracted Large Resource Data Type Tag item. +*/ +static inline u16 pci_vpd_lrdt_tag (const u8 *lrdt) +{ +return (u16)(lrdt[0 ] &PCI_VPD_LRDT_TIN_MASK ); +} + +/** * pci_vpd_srdt_size - Extracts the Small Resource Data Type length * @lrdt: Pointer to the beginning of the Small Resource Data Type tag *
@@ -1867,6 +1879,17 @@ static inline u8 pci_vpd_srdt_size(const u8 *srdt) }/** +* pci_vpd_srdt_tag - Extracts the Small Resource Data Type Tag Item +* @lrdt: Pointer to the beginning of the Small Resource Data Type tag +* +* Returns the extracted Small Resource Data Type Tag Item. +*/ +static inline u8 pci_vpd_srdt_tag (const u8 *srdt) +{ +return ((*srdt) &PCI_VPD_SRDT_TIN_MASK ) >>3 ; +} + +/** * pci_vpd_info_field_size - Extracts the information field length * @lrdt: Pointer to the beginning of an information field header *
Signature · 1 line
1.8.5.6
PATCH v4 2/2 pci: Update VPD size with correct length
In reply to Hannes ReineckePCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.
--- drivers/pci/access.c | 81 ++++++++++++++++++++++++++++++++++++++++++++++++++++ drivers/pci/pci.h | 1 + 2 files changed, 82 insertions(+)
drivers/pci/access.c +81 -0
@@ -206,6 +206,18 @@ ssize_t pci_write_vpd(struct pci_dev *dev, loff_t pos, size_t count, const void }EXPORT_SYMBOL (pci_write_vpd); +/** +* pci_vpd_size - Determine the size of the Vital Product Data +* @dev: pci device struct +* +*/ +size_t pci_vpd_size (struct pci_dev *dev) +{ +if (!dev->vpd || !dev->vpd->ops) +return 0 ; +return dev->vpd->ops->size (dev); +} +/* * The following routines are to prevent the user from accessing PCI config * space when it's unsafe to do so. Some devices require this during BIST and
@@ -428,6 +440,53 @@ out:return ret ? ret : count; } +/** +* pci_vpd_size - determine actual size of Vital Product Data +* @dev: pci device struct +* +*/ +static size_t pci_vpd_pci22_size (struct pci_dev *dev) +{ +ssize_t off =0 ; +unsigned char header[1 +2 ];/* 1 byte tag, 2 bytes length */ + +while (pci_read_vpd (dev, off,1 , header) ==1 ) { +unsigned char tag; + +if (header[0 ] &PCI_VPD_LRDT ) { +/* Large Resource Data Type Tag */ + tag =pci_vpd_lrdt_tag (header); +/* Only read length from known tag items */ +if ((tag ==PCI_VPD_LTIN_ID_STRING ) || + (tag ==PCI_VPD_LTIN_RO_DATA ) || + (tag ==PCI_VPD_LTIN_RW_DATA )) { +if (pci_read_vpd (dev, off+1 ,2 , + &header[1 ]) !=2 ) +return off +1 ; + off +=PCI_VPD_LRDT_TAG_SIZE + +pci_vpd_lrdt_size (header); + } + }else { +/* Short Resource Data Type Tag */ + off +=PCI_VPD_SRDT_TAG_SIZE + +pci_vpd_srdt_size (header); + tag =pci_vpd_srdt_tag (header); + } +if (tag ==PCI_VPD_STIN_END )/* End tag descriptor */ +return off; +if ((tag !=PCI_VPD_LTIN_ID_STRING ) && + (tag !=PCI_VPD_LTIN_RO_DATA ) && + (tag !=PCI_VPD_LTIN_RW_DATA )) { +dev_dbg (&dev->dev, +"invalid %s vpd tag %02x at offset %zu." , + (header[0 ] &PCI_VPD_LRDT ) ?"large" :"short" , + tag, off); +break ; + } + } +return 0 ; +} +static void pci_vpd_pci22_release (struct pci_dev *dev) { kfree(container_of(dev->vpd, struct pci_vpd_pci22, base));
@@ -436,6 +495,7 @@ static void pci_vpd_pci22_release(struct pci_dev *dev)static const struct pci_vpd_ops pci_vpd_pci22_ops = { .read = pci_vpd_pci22_read, .write = pci_vpd_pci22_write, + .size = pci_vpd_pci22_size, .release = pci_vpd_pci22_release, };
@@ -469,9 +529,29 @@ static ssize_t pci_vpd_f0_write(struct pci_dev *dev, loff_t pos, size_t count,return ret; } +static size_t pci_vpd_f0_size (struct pci_dev *dev) +{ +struct pci_dev *tdev =pci_get_slot (dev->bus, +PCI_DEVFN (PCI_SLOT (dev->devfn),0 )); +ssize_t len =0 ; + +if (!tdev) +return 0 ; + +if (tdev->vpd) { +struct pci_vpd_pci22 *vpd = + container_of(dev->vpd, struct pci_vpd_pci22, base); + + len = vpd->base.len; + } +pci_dev_put (tdev); +return len; +} +static const struct pci_vpd_ops pci_vpd_f0_ops = { .read = pci_vpd_f0_read, .write = pci_vpd_f0_write, + .size = pci_vpd_f0_size, .release = pci_vpd_pci22_release, };
@@ -497,6 +577,7 @@ int pci_vpd_pci22_init(struct pci_dev *dev) vpd->cap = cap; vpd->busy = false; dev->vpd = &vpd->base; + vpd->base.len =pci_vpd_size (dev);return 0 ; }
drivers/pci/pci.h +1 -0
@@ -97,6 +97,7 @@ static inline bool pci_has_subordinate(struct pci_dev *pci_dev)struct pci_vpd_ops {ssize_t (*read)(struct pci_dev *dev,loff_t pos,size_t count,void *buf);ssize_t (*write)(struct pci_dev *dev,loff_t pos,size_t count,const void *buf); +size_t (*size)(struct pci_dev *dev);void (*release)(struct pci_dev *dev); };
Signature · 1 line
1.8.5.6
Re: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 9 lines · Hannes Reinecke
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.
This isn't what I had in mind. There is no need to add an f0 version of the size function. The size for all functions other than function 0 when the F0 flag is set is 0. We aren't going to be reading their VPD, we only read the VPD region of function 0.
If you just moved the call to pci_vpd_size into the conditional statement in the init function you could then set the length to 0 in the branch taken for if the F0 flag is set. You might have to reorder the init function a bit but there is no need to add additional function pointers.
- Alex
Re: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 11 lines · Alexander Duyck
Quoted 9 lines · Hannes Reinecke
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.
Quoted 5 lines
This isn't what I had in mind. There is no need to add an f0 version of the size function. The size for all functions other than function 0 when the F0 flag is set is 0. We aren't going to be reading their VPD, we only read the VPD region of function 0.
Ah. (I'm a bit confused about the proposed action for VPD other than function 0).
So the idea here is to _disallow_ access to VPDs from functions other than '0' unless these functions have different PCI IDs?
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke zSeries & Storage [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG Nürnberg)
Re: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 20 lines · Hannes Reinecke
Quoted 14 lines · Alexander Duyck
Quoted 10 lines · Hannes Reinecke
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.Quoted 5 lines
This isn't what I had in mind. There is no need to add an f0 version of the size function. The size for all functions other than function 0 when the F0 flag is set is 0. We aren't going to be reading their VPD, we only read the VPD region of function 0.
Quoted 4 lines
Ah. (I'm a bit confused about the proposed action for VPD other than function 0).
So the idea here is to _disallow_ access to VPDs from functions other than '0' unless these functions have different PCI IDs?
If you take a look at the F0 functions what they do is bypass the VPD of the functions other than function 0. As such setting the size to 0 should really have no effect since the VPD of the function isn't actually read if the F0 flag is set.
- Alex
Re: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 26 lines · Alexander Duyck
Quoted 20 lines · Hannes Reinecke
Quoted 14 lines · Alexander Duyck
Quoted 10 lines · Hannes Reinecke
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.Quoted 5 lines
This isn't what I had in mind. There is no need to add an f0 version of the size function. The size for all functions other than function 0 when the F0 flag is set is 0. We aren't going to be reading their VPD, we only read the VPD region of function 0.
Quoted 4 lines
Ah. (I'm a bit confused about the proposed action for VPD other than function 0).
So the idea here is to _disallow_ access to VPDs from functions other than '0' unless these functions have different PCI IDs?
Quoted 5 lines
If you take a look at the F0 functions what they do is bypass the VPD of the functions other than function 0. As such setting the size to 0 should really have no effect since the VPD of the function isn't actually read if the F0 flag is set.
Setting the size to '0' effectively inhibits you to read the VPD data. So if we were to return '0' for PCI devices with the F0 bit set we will never ever to be able to read (or write) _any_ VPD data for that PCI device/function. Which would be rendering all these F0 accessors pointless, and we might as well remove them.
Hence my confusion.
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke zSeries & Storage [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG Nürnberg)
RE: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 32 lines · Alexander Duyck
Quoted 26 lines · Hannes Reinecke
Quoted 20 lines · Alexander Duyck
Quoted 14 lines · Hannes Reinecke
Quoted 10 lines
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.Quoted 5 lines
This isn't what I had in mind. There is no need to add an f0 version of the size function. The size for all functions other than function 0 when the F0 flag is set is 0. We aren't going to be reading their VPD, we only read the VPD region of function 0.
Quoted 4 lines
Ah. (I'm a bit confused about the proposed action for VPD other than function 0).
So the idea here is to _disallow_ access to VPDs from functions other than '0' unless these functions have different PCI IDs?Quoted 5 lines
If you take a look at the F0 functions what they do is bypass the VPD of the functions other than function 0. As such setting the size to 0 should really have no effect since the VPD of the function isn't actually read if the F0 flag is set.
Quoted 7 lines
Setting the size to '0' effectively inhibits you to read the VPD data. So if we were to return '0' for PCI devices with the F0 bit set we will never ever to be able to read (or write) _any_ VPD data for that PCI device/function.
Which would be rendering all these F0 accessors pointless, and we might as well remove them.
I had posted a patch recently to enable exposing the VPD-R valyes to sysfs. I need access to these to parse into systemd for network naming (biosdevname style names).
The VPD-R is a readonly area contained within the PCI Vital Product data. There are some standard and vendor-specific keys stored in this region.
PN = Part Number
SN = Serial Number
MN = Manufacturer ID
Vx = Vendor-specific (x=0..9 A..Z)
Biosdevname/Systemd will use these VPD keys for determining Network partitioning and port numbers for NIC cards
--- drivers/pci/pci-sysfs.c | 216 +++++++++++++++++++++++++++++++++++++++++++++++ include/linux/pci.h | 2 + 2 files changed, 218 insertions(+), 0 deletions(-)
drivers/pci/pci-sysfs.c +216 -0
@@ -1295,6 +1295,210 @@ static struct bin_attribute pcie_config_attr = { .write = pci_write_config, }; +static umode_t vpd_attr_exist (struct kobject *kobj, +struct attribute *attr,int n) +{ +struct device *dev; +struct pci_dev *pdev; +const char *name; +int i; + + dev = container_of(kobj, struct device, kobj); + pdev =to_pci_dev (dev); + + name = attr->name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i =pci_vpd_find_info_keyword (pdev->vpdr_data,0 , + pdev->vpdr_len, name); +return (i >=0 ?S_IRUGO :0 ); +} + +static ssize_t vpd_attr_show (struct device *dev, +struct device_attribute *attr,char *buf) +{ +struct pci_dev *pdev; +const char *name; +char kv_data[257 ] = {0 }; +int i, len; + + pdev =to_pci_dev (dev); + name = attr->attr.name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i =pci_vpd_find_info_keyword (pdev->vpdr_data,0 , + pdev->vpdr_len, name); +if (i >=0 ) { + len =pci_vpd_info_field_size (&pdev->vpdr_data[i]); +memcpy (kv_data, pdev->vpdr_data + i + +PCI_VPD_INFO_FLD_HDR_SIZE , len); +return scnprintf (buf,PAGE_SIZE ,"%s\n" , kv_data); + } +return 0 ; +} + +#define VPD_ATTR_RO (x) \ +struct device_attribute vpd ## x = { \ + .attr = { .name = #x, .mode = S_IRUGO | S_IWUSR }, \ + .show = vpd_attr_show \ +} + +VPD_ATTR_RO (PN ); +VPD_ATTR_RO (EC ); +VPD_ATTR_RO (MN ); +VPD_ATTR_RO (SN ); +VPD_ATTR_RO (V0 ); +VPD_ATTR_RO (V1 ); +VPD_ATTR_RO (V2 ); +VPD_ATTR_RO (V3 ); +VPD_ATTR_RO (V4 ); +VPD_ATTR_RO (V5 ); +VPD_ATTR_RO (V6 ); +VPD_ATTR_RO (V7 ); +VPD_ATTR_RO (V8 ); +VPD_ATTR_RO (V9 ); +VPD_ATTR_RO (VA ); +VPD_ATTR_RO (VB ); +VPD_ATTR_RO (VC ); +VPD_ATTR_RO (VD ); +VPD_ATTR_RO (VE ); +VPD_ATTR_RO (VF ); +VPD_ATTR_RO (VG ); +VPD_ATTR_RO (VH ); +VPD_ATTR_RO (VI ); +VPD_ATTR_RO (VJ ); +VPD_ATTR_RO (VK ); +VPD_ATTR_RO (VL ); +VPD_ATTR_RO (VM ); +VPD_ATTR_RO (VN ); +VPD_ATTR_RO (VO ); +VPD_ATTR_RO (VP ); +VPD_ATTR_RO (VQ ); +VPD_ATTR_RO (VR ); +VPD_ATTR_RO (VS ); +VPD_ATTR_RO (VT ); +VPD_ATTR_RO (VU ); +VPD_ATTR_RO (VV ); +VPD_ATTR_RO (VW ); +VPD_ATTR_RO (VX ); +VPD_ATTR_RO (VY ); +VPD_ATTR_RO (VZ ); + +static struct attribute *vpd_attributes[] = { + &vpdPN.attr, + &vpdEC.attr, + &vpdMN.attr, + &vpdSN.attr, + &vpdV0.attr, + &vpdV1.attr, + &vpdV2.attr, + &vpdV3.attr, + &vpdV4.attr, + &vpdV5.attr, + &vpdV6.attr, + &vpdV7.attr, + &vpdV8.attr, + &vpdV9.attr, + &vpdVA.attr, + &vpdVB.attr, + &vpdVC.attr, + &vpdVD.attr, + &vpdVE.attr, + &vpdVF.attr, + &vpdVG.attr, + &vpdVH.attr, + &vpdVI.attr, + &vpdVJ.attr, + &vpdVK.attr, + &vpdVL.attr, + &vpdVM.attr, + &vpdVN.attr, + &vpdVO.attr, + &vpdVP.attr, + &vpdVQ.attr, + &vpdVR.attr, + &vpdVS.attr, + &vpdVT.attr, + &vpdVU.attr, + &vpdVV.attr, + &vpdVW.attr, + &vpdVX.attr, + &vpdVY.attr, + &vpdVZ.attr, +NULL , +}; + +static struct attribute_group vpd_attr_group = { + .name ="vpdr" , + .attrs = vpd_attributes, + .is_visible = vpd_attr_exist, +}; + + +static int pci_get_vpd_tag (struct pci_dev *dev,int off,int *len) +{ +u8 tag[3 ]; +int rc, tlen; + + *len =0 ; +/* Quirk Atheros cards, reading VPD hangs system for 20s */ +if (dev->vendor ==PCI_VENDOR_ID_ATHEROS || + dev->vendor ==PCI_VENDOR_ID_ATTANSIC ) +return -ENOENT ; + rc =pci_read_vpd (dev, off,1 , tag); +if (rc !=1 ) +return -ENOENT ; +if (tag[0 ] ==0x00 || tag[0 ] ==0xFF || tag[0 ] ==0x7F ) +return -ENOENT ; +if (tag[0 ] &PCI_VPD_LRDT ) { + rc =pci_read_vpd (dev, off+1 ,2 , tag+1 ); +if (rc !=2 ) +return -ENOENT ; + tlen =pci_vpd_lrdt_size (tag) + +PCI_VPD_LRDT_TAG_SIZE ; + }else { + tlen =pci_vpd_srdt_size (tag) + +PCI_VPD_SRDT_TAG_SIZE ; + tag[0 ] &= ~PCI_VPD_SRDT_LEN_MASK ; + } +/* Verify VPD tag fits in area */ +if (tlen + off > dev->vpd->len) +return -ENOENT ; + *len = tlen; +return tag[0 ]; +} + +static int pci_load_vpdr (struct pci_dev *dev) +{ +int rlen, ilen, tag, rc; + +/* Check for VPD-I and VPD-R tag */ + tag =pci_get_vpd_tag (dev,0 , &ilen); +if (tag !=PCI_VPD_LRDT_ID_STRING ) +return -ENOENT ; + tag =pci_get_vpd_tag (dev, ilen, &rlen); +if (tag !=PCI_VPD_LRDT_RO_DATA ) +return -ENOENT ; + + rlen -=PCI_VPD_LRDT_TAG_SIZE ; + dev->vpdr_len = rlen; + dev->vpdr_data =kzalloc (rlen,GFP_ATOMIC ); +if (dev->vpdr_data ==NULL ) +return -ENOMEM ; + + rc =pci_read_vpd (dev, ilen +PCI_VPD_LRDT_TAG_SIZE , + rlen, dev->vpdr_data); +if (rc != rlen) + goto error; +if (sysfs_create_group (&dev->dev.kobj, &vpd_attr_group)) + goto error; +return 0 ; + error: +kfree (dev->vpdr_data); + dev->vpdr_len =0 ; +return -ENOENT ; +} +static ssize_t reset_store (struct device *dev,struct device_attribute *attr,const char *buf,size_t count) {
@@ -1340,6 +1544,8 @@ static int pci_create_capabilities_sysfs(struct pci_dev *dev)return retval; } dev->vpd->attr = attr; + +pci_load_vpdr (dev); }/* Active State Power Management */
@@ -1356,6 +1562,11 @@ static int pci_create_capabilities_sysfs(struct pci_dev *dev) error:pcie_aspm_remove_sysfs_dev_files (dev);if (dev->vpd && dev->vpd->attr) { +if (dev->vpdr_data) { +sysfs_remove_group (&dev->dev.kobj, &vpd_attr_group); +kfree (dev->vpdr_data); + dev->vpdr_data =NULL ; + }sysfs_remove_bin_file (&dev->dev.kobj, dev->vpd->attr);kfree (dev->vpd->attr); }
@@ -1438,6 +1649,11 @@ err:static void pci_remove_capabilities_sysfs (struct pci_dev *dev) {if (dev->vpd && dev->vpd->attr) { +if (dev->vpdr_data) { +sysfs_remove_group (&dev->dev.kobj, &vpd_attr_group); +kfree (dev->vpdr_data); + dev->vpdr_data =NULL ; + }sysfs_remove_bin_file (&dev->dev.kobj, dev->vpd->attr);kfree (dev->vpd->attr); }
include/linux/pci.h +2 -0
@@ -372,6 +372,8 @@ struct pci_dev {const struct attribute_group **msi_irq_groups;#endif struct pci_vpd *vpd; +int vpdr_len; +u8 *vpdr_data; #ifdef CONFIG_PCI_ATS union { struct pci_sriov *sriov; /* SR-IOV capability related */
Signature · 1 line
1.7.1
Quoted 16 lines
Hence my confusion.
Cheers,
Hannes
--
Dr. Hannes Reinecke zSeries & Storage
[email protected] +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG Nürnberg)
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 48 lines · On Mon, Dec 28, 2015 at 9:29 PM, <[email protected]> wrote:
Quoted 32 lines · Alexander Duyck
Quoted 26 lines · Hannes Reinecke
Quoted 20 lines · Alexander Duyck
Quoted 14 lines · Hannes Reinecke
Quoted 10 lines
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be smaller than that. To figure out the actual size one has to read the VPD area until the 'end marker' is reached.
Trying to read VPD data beyond that marker results in 'interesting' effects, from simple read errors to crashing the card. And to make matters worse not every PCI card implements this properly, leaving us with no 'end' marker or even completely invalid data. This path modifies the size of the VPD attribute to the available size, or set it to '0' if no valid data could be read.Quoted 5 lines
This isn't what I had in mind. There is no need to add an f0 version of the size function. The size for all functions other than function 0 when the F0 flag is set is 0. We aren't going to be reading their VPD, we only read the VPD region of function 0.
Quoted 4 lines
Ah. (I'm a bit confused about the proposed action for VPD other than function 0).
So the idea here is to _disallow_ access to VPDs from functions other than '0' unless these functions have different PCI IDs?Quoted 5 lines
If you take a look at the F0 functions what they do is bypass the VPD of the functions other than function 0. As such setting the size to 0 should really have no effect since the VPD of the function isn't actually read if the F0 flag is set.
Quoted 7 lines
Setting the size to '0' effectively inhibits you to read the VPD data. So if we were to return '0' for PCI devices with the F0 bit set we will never ever to be able to read (or write) _any_ VPD data for that PCI device/function.
Which would be rendering all these F0 accessors pointless, and we might as well remove them.I had posted a patch recently to enable exposing the VPD-R valyes to sysfs. I need access to these to parse into systemd for network naming (biosdevname style names).
The VPD-R is a readonly area contained within the PCI Vital Product data. There are some standard and vendor-specific keys stored in this region.
Quoted 7 lines
PN = Part Number
SN = Serial Number
MN = Manufacturer ID
Vx = Vendor-specific (x=0..9 A..Z)Biosdevname/Systemd will use these VPD keys for determining Network partitioning and port numbers for NIC cards
Can you please repost this as a patch instead of as a reply to our thread about VPD size. The fact is the subject is misleading as your patch isn't actually related to VPD sizing.
Quoted 23 lines
Signed-off-by Jordan Hargrave <[email protected]>--- drivers/pci/pci-sysfs.c | 216 +++++++++++++++++++++++++++++++++++++++++++++++ include/linux/pci.h | 2 + 2 files changed, 218 insertions(+), 0 deletions(-)drivers/pci/pci-sysfs.c +9
-0@@ -1295,6 +1295,210 @@ static struct bin_attribute pcie_config_attr = { .write = pci_write_config, }; +static umode_t vpd_attr_exist (struct kobject *kobj, +struct attribute *attr,int n) +{ +struct device *dev; +struct pci_dev *pdev; +const char *name; +int i; + + dev = container_of(kobj, struct device, kobj);
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1316
Quoted 10 lines
drivers/pci/pci-sysfs.c +10
-0+ pdev =to_pci_dev (dev); + + name = attr->name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i =pci_vpd_find_info_keyword (pdev->vpdr_data,0 , + pdev->vpdr_len, name); +return (i >=0 ?S_IRUGO :0 ); +} +
So I assume there is another patch that implements pci_vpd_find_info_keyword so that it can go through the vpdr_data and parse it?
Quoted 121 lines
drivers/pci/pci-sysfs.c +121
-0+static ssize_t vpd_attr_show (struct device *dev, +struct device_attribute *attr,char *buf) +{ +struct pci_dev *pdev; +const char *name; +char kv_data[257 ] = {0 }; +int i, len; + + pdev =to_pci_dev (dev); + name = attr->attr.name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i =pci_vpd_find_info_keyword (pdev->vpdr_data,0 , + pdev->vpdr_len, name); +if (i >=0 ) { + len =pci_vpd_info_field_size (&pdev->vpdr_data[i]); +memcpy (kv_data, pdev->vpdr_data + i + +PCI_VPD_INFO_FLD_HDR_SIZE , len); +return scnprintf (buf,PAGE_SIZE ,"%s\n" , kv_data); + } +return 0 ; +} + +#define VPD_ATTR_RO (x) \ +struct device_attribute vpd ## x = { \ + .attr = { .name = #x, .mode = S_IRUGO | S_IWUSR }, \ + .show = vpd_attr_show \ +} + +VPD_ATTR_RO (PN ); +VPD_ATTR_RO (EC ); +VPD_ATTR_RO (MN ); +VPD_ATTR_RO (SN ); +VPD_ATTR_RO (V0 ); +VPD_ATTR_RO (V1 ); +VPD_ATTR_RO (V2 ); +VPD_ATTR_RO (V3 ); +VPD_ATTR_RO (V4 ); +VPD_ATTR_RO (V5 ); +VPD_ATTR_RO (V6 ); +VPD_ATTR_RO (V7 ); +VPD_ATTR_RO (V8 ); +VPD_ATTR_RO (V9 ); +VPD_ATTR_RO (VA ); +VPD_ATTR_RO (VB ); +VPD_ATTR_RO (VC ); +VPD_ATTR_RO (VD ); +VPD_ATTR_RO (VE ); +VPD_ATTR_RO (VF ); +VPD_ATTR_RO (VG ); +VPD_ATTR_RO (VH ); +VPD_ATTR_RO (VI ); +VPD_ATTR_RO (VJ ); +VPD_ATTR_RO (VK ); +VPD_ATTR_RO (VL ); +VPD_ATTR_RO (VM ); +VPD_ATTR_RO (VN ); +VPD_ATTR_RO (VO ); +VPD_ATTR_RO (VP ); +VPD_ATTR_RO (VQ ); +VPD_ATTR_RO (VR ); +VPD_ATTR_RO (VS ); +VPD_ATTR_RO (VT ); +VPD_ATTR_RO (VU ); +VPD_ATTR_RO (VV ); +VPD_ATTR_RO (VW ); +VPD_ATTR_RO (VX ); +VPD_ATTR_RO (VY ); +VPD_ATTR_RO (VZ ); + +static struct attribute *vpd_attributes[] = { + &vpdPN.attr, + &vpdEC.attr, + &vpdMN.attr, + &vpdSN.attr, + &vpdV0.attr, + &vpdV1.attr, + &vpdV2.attr, + &vpdV3.attr, + &vpdV4.attr, + &vpdV5.attr, + &vpdV6.attr, + &vpdV7.attr, + &vpdV8.attr, + &vpdV9.attr, + &vpdVA.attr, + &vpdVB.attr, + &vpdVC.attr, + &vpdVD.attr, + &vpdVE.attr, + &vpdVF.attr, + &vpdVG.attr, + &vpdVH.attr, + &vpdVI.attr, + &vpdVJ.attr, + &vpdVK.attr, + &vpdVL.attr, + &vpdVM.attr, + &vpdVN.attr, + &vpdVO.attr, + &vpdVP.attr, + &vpdVQ.attr, + &vpdVR.attr, + &vpdVS.attr, + &vpdVT.attr, + &vpdVU.attr, + &vpdVV.attr, + &vpdVW.attr, + &vpdVX.attr, + &vpdVY.attr, + &vpdVZ.attr, +NULL , +}; + +static struct attribute_group vpd_attr_group = { + .name ="vpdr" , + .attrs = vpd_attributes, + .is_visible = vpd_attr_exist, +}; + +
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1447
Quoted 10 lines
drivers/pci/pci-sysfs.c +10
-0+static int pci_get_vpd_tag (struct pci_dev *dev,int off,int *len) +{ +u8 tag[3 ]; +int rc, tlen; + + *len =0 ; +/* Quirk Atheros cards, reading VPD hangs system for 20s */ +if (dev->vendor ==PCI_VENDOR_ID_ATHEROS || + dev->vendor ==PCI_VENDOR_ID_ATTANSIC ) +return -ENOENT ;
I'm not really sure this is the right place for this type of quirk. If this is really an issue maybe we should just disable VPD for these devices. Otherwise there isn't anything to stop someone from going in and reading the VPD region via the existing VPD interfaces.
Quoted 31 lines
drivers/pci/pci-sysfs.c +31
-0+ rc =pci_read_vpd (dev, off,1 , tag); +if (rc !=1 ) +return -ENOENT ; +if (tag[0 ] ==0x00 || tag[0 ] ==0xFF || tag[0 ] ==0x7F ) +return -ENOENT ; +if (tag[0 ] &PCI_VPD_LRDT ) { + rc =pci_read_vpd (dev, off+1 ,2 , tag+1 ); +if (rc !=2 ) +return -ENOENT ; + tlen =pci_vpd_lrdt_size (tag) + +PCI_VPD_LRDT_TAG_SIZE ; + }else { + tlen =pci_vpd_srdt_size (tag) + +PCI_VPD_SRDT_TAG_SIZE ; + tag[0 ] &= ~PCI_VPD_SRDT_LEN_MASK ; + } +/* Verify VPD tag fits in area */ +if (tlen + off > dev->vpd->len) +return -ENOENT ; + *len = tlen; +return tag[0 ]; +} + +static int pci_load_vpdr (struct pci_dev *dev) +{ +int rlen, ilen, tag, rc; + +/* Check for VPD-I and VPD-R tag */ + tag =pci_get_vpd_tag (dev,0 , &ilen); +if (tag !=PCI_VPD_LRDT_ID_STRING ) +return -ENOENT ;
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1488
Quoted 10 lines
drivers/pci/pci-sysfs.c +10
-0+ tag =pci_get_vpd_tag (dev, ilen, &rlen); +if (tag !=PCI_VPD_LRDT_RO_DATA ) +return -ENOENT ; + + rlen -=PCI_VPD_LRDT_TAG_SIZE ; + dev->vpdr_len = rlen; + dev->vpdr_data =kzalloc (rlen,GFP_ATOMIC ); +if (dev->vpdr_data ==NULL ) +return -ENOMEM ; +
Why not cache the ID string as well? Seems like it might be a field people would want to read on a regular basis in order to find out what is there.
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1501
Quoted 13 lines
drivers/pci/pci-sysfs.c +13
-0+ rc =pci_read_vpd (dev, ilen +PCI_VPD_LRDT_TAG_SIZE , + rlen, dev->vpdr_data); +if (rc != rlen) + goto error; +if (sysfs_create_group (&dev->dev.kobj, &vpd_attr_group)) + goto error; +return 0 ; + error: +kfree (dev->vpdr_data); + dev->vpdr_len =0 ; +return -ENOENT ; +} +
This bit here needs to reset vpdr_data back to NULL. Otherwise you could cause memory corruption via a double free in your two clean-up routines called out below.
Quoted 41 lines
drivers/pci/pci-sysfs.c +12
-0static ssize_t reset_store (struct device *dev,struct device_attribute *attr,const char *buf,size_t count) {@@ -1340,6 +1544,8 @@ static int pci_create_capabilities_sysfs(struct pci_dev *dev)return retval; } dev->vpd->attr = attr; + +pci_load_vpdr (dev); }/* Active State Power Management */ @@ -1356,6 +1562,11 @@ static int pci_create_capabilities_sysfs(struct pci_dev *dev) error:pcie_aspm_remove_sysfs_dev_files (dev);if (dev->vpd && dev->vpd->attr) { +if (dev->vpdr_data) { +sysfs_remove_group (&dev->dev.kobj, &vpd_attr_group); +kfree (dev->vpdr_data); + dev->vpdr_data =NULL ; + }sysfs_remove_bin_file (&dev->dev.kobj, dev->vpd->attr);kfree (dev->vpd->attr); }@@ -1438,6 +1649,11 @@ err:static void pci_remove_capabilities_sysfs (struct pci_dev *dev) {if (dev->vpd && dev->vpd->attr) { +if (dev->vpdr_data) { +sysfs_remove_group (&dev->dev.kobj, &vpd_attr_group); +kfree (dev->vpdr_data); + dev->vpdr_data =NULL ; + }sysfs_remove_bin_file (&dev->dev.kobj, dev->vpd->attr);kfree (dev->vpd->attr); }include/linux/pci.h +0
-0
include/linux/pci.h struct pci_dev { · line 379
Quoted 8 lines
include/linux/pci.h +2
-0@@ -372,6 +372,8 @@ struct pci_dev {const struct attribute_group **msi_irq_groups;#endif struct pci_vpd *vpd; +int vpdr_len; +u8 *vpdr_data; #ifdef CONFIG_PCI_ATS union { struct pci_sriov *sriov; /* SR-IOV capability related */
Instead of placing vpdr_len outside of the data why not just leave it in there? It seems like you could then make use of some of the same parsing logic regardless of if you are working with the cached copy or the actual VPD.
RE: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 14 lines · On Mon, Dec 28, 2015 at 9:29 PM, <[email protected]> wrote:
Quoted 13 lines
I had posted a patch recently to enable exposing the VPD-R valyes to sysfs. I need access to these to parse into systemd for network naming (biosdevname style names).
The VPD-R is a readonly area contained within the PCI Vital Product data. There are some standard and vendor-specific keys stored in this region.
PN = Part Number
SN = Serial Number
MN = Manufacturer ID
Vx = Vendor-specific (x=0..9 A..Z)
Quoted 7 lines
Quoted 2 lines
Biosdevname/Systemd will use these VPD keys for determining Network partitioning and port numbers for NIC cards
Can you please repost this as a patch instead of as a reply to our thread about VPD size. The fact is the subject is misleading as your patch isn't actually related to VPD sizing.
I had already posted this a few weeks back but never got any feedback.
[PATCH] Create sysfs entries for VPD-R keys
https://marc.info/?l=linux-pci&m=144959803316031&w=2
Quoted 29 lines
Quoted 29 lines
Signed-off-by Jordan Hargrave <[email protected]>--- drivers/pci/pci-sysfs.c | 216 +++++++++++++++++++++++++++++++++++++++++++++++ include/linux/pci.h | 2 + 2 files changed, 218 insertions(+), 0 deletions(-)drivers/pci/pci-sysfs.c +15
-0@@ -1295,6 +1295,210 @@ static struct bin_attribute pcie_config_attr = { .write = pci_write_config, }; +static umode_t vpd_attr_exist (struct kobject *kobj, +struct attribute *attr,int n) +{ +struct device *dev; +struct pci_dev *pdev; +const char *name; +int i; + + dev = container_of(kobj, struct device, kobj); + pdev =to_pci_dev (dev); + + name = attr->name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i = pci_vpd_find_info_keyword(pdev->vpdr_data, 0,
Quoted 9 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1316
Quoted 4 lines
drivers/pci/pci-sysfs.c +4
-0+ pdev->vpdr_len, name); +return (i >=0 ?S_IRUGO :0 ); +} +So I assume there is another patch that implements
pci_vpd_find_info_keywordso that it can go through thevpdr_dataand parse it?
That's already an existing function in drivers/pci/vpd.c
Quoted 128 lines
Quoted 128 lines
drivers/pci/pci-sysfs.c +128
-0+static ssize_t vpd_attr_show (struct device *dev, +struct device_attribute *attr,char *buf) +{ +struct pci_dev *pdev; +const char *name; +char kv_data[257 ] = {0 }; +int i, len; + + pdev =to_pci_dev (dev); + name = attr->attr.name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i =pci_vpd_find_info_keyword (pdev->vpdr_data,0 , + pdev->vpdr_len, name); +if (i >=0 ) { + len =pci_vpd_info_field_size (&pdev->vpdr_data[i]); +memcpy (kv_data, pdev->vpdr_data + i + +PCI_VPD_INFO_FLD_HDR_SIZE , len); +return scnprintf (buf,PAGE_SIZE ,"%s\n" , kv_data); + } +return 0 ; +} + +#define VPD_ATTR_RO (x) \ +struct device_attribute vpd ## x = { \ + .attr = { .name = #x, .mode = S_IRUGO | S_IWUSR }, \ + .show = vpd_attr_show \ +} + +VPD_ATTR_RO (PN ); +VPD_ATTR_RO (EC ); +VPD_ATTR_RO (MN ); +VPD_ATTR_RO (SN ); +VPD_ATTR_RO (V0 ); +VPD_ATTR_RO (V1 ); +VPD_ATTR_RO (V2 ); +VPD_ATTR_RO (V3 ); +VPD_ATTR_RO (V4 ); +VPD_ATTR_RO (V5 ); +VPD_ATTR_RO (V6 ); +VPD_ATTR_RO (V7 ); +VPD_ATTR_RO (V8 ); +VPD_ATTR_RO (V9 ); +VPD_ATTR_RO (VA ); +VPD_ATTR_RO (VB ); +VPD_ATTR_RO (VC ); +VPD_ATTR_RO (VD ); +VPD_ATTR_RO (VE ); +VPD_ATTR_RO (VF ); +VPD_ATTR_RO (VG ); +VPD_ATTR_RO (VH ); +VPD_ATTR_RO (VI ); +VPD_ATTR_RO (VJ ); +VPD_ATTR_RO (VK ); +VPD_ATTR_RO (VL ); +VPD_ATTR_RO (VM ); +VPD_ATTR_RO (VN ); +VPD_ATTR_RO (VO ); +VPD_ATTR_RO (VP ); +VPD_ATTR_RO (VQ ); +VPD_ATTR_RO (VR ); +VPD_ATTR_RO (VS ); +VPD_ATTR_RO (VT ); +VPD_ATTR_RO (VU ); +VPD_ATTR_RO (VV ); +VPD_ATTR_RO (VW ); +VPD_ATTR_RO (VX ); +VPD_ATTR_RO (VY ); +VPD_ATTR_RO (VZ ); + +static struct attribute *vpd_attributes[] = { + &vpdPN.attr, + &vpdEC.attr, + &vpdMN.attr, + &vpdSN.attr, + &vpdV0.attr, + &vpdV1.attr, + &vpdV2.attr, + &vpdV3.attr, + &vpdV4.attr, + &vpdV5.attr, + &vpdV6.attr, + &vpdV7.attr, + &vpdV8.attr, + &vpdV9.attr, + &vpdVA.attr, + &vpdVB.attr, + &vpdVC.attr, + &vpdVD.attr, + &vpdVE.attr, + &vpdVF.attr, + &vpdVG.attr, + &vpdVH.attr, + &vpdVI.attr, + &vpdVJ.attr, + &vpdVK.attr, + &vpdVL.attr, + &vpdVM.attr, + &vpdVN.attr, + &vpdVO.attr, + &vpdVP.attr, + &vpdVQ.attr, + &vpdVR.attr, + &vpdVS.attr, + &vpdVT.attr, + &vpdVU.attr, + &vpdVV.attr, + &vpdVW.attr, + &vpdVX.attr, + &vpdVY.attr, + &vpdVZ.attr, +NULL , +}; + +static struct attribute_group vpd_attr_group = { + .name ="vpdr" , + .attrs = vpd_attributes, + .is_visible = vpd_attr_exist, +}; + + +static int pci_get_vpd_tag (struct pci_dev *dev,int off,int *len) +{ +u8 tag[3 ]; +int rc, tlen; + + *len =0 ; +/* Quirk Atheros cards, reading VPD hangs system for 20s */
Quoted 9 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1447
Quoted 3 lines
drivers/pci/pci-sysfs.c +3
-0+if (dev->vendor ==PCI_VENDOR_ID_ATHEROS || + dev->vendor ==PCI_VENDOR_ID_ATTANSIC ) +return -ENOENT ;I'm not really sure this is the right place for this type of quirk. If this is really an issue maybe we should just disable VPD for these devices. Otherwise there isn't anything to stop someone from going in and reading the VPD region via the existing VPD interfaces.
This could be moved elsewhere. There's a similar quirk for megaraid cards that was posted recently. [PATCH v4] pci: Limit VPD length for megaraid_sas adapter
Quoted 37 lines
Quoted 37 lines
drivers/pci/pci-sysfs.c +37
-0+ rc =pci_read_vpd (dev, off,1 , tag); +if (rc !=1 ) +return -ENOENT ; +if (tag[0 ] ==0x00 || tag[0 ] ==0xFF || tag[0 ] ==0x7F ) +return -ENOENT ; +if (tag[0 ] &PCI_VPD_LRDT ) { + rc =pci_read_vpd (dev, off+1 ,2 , tag+1 ); +if (rc !=2 ) +return -ENOENT ; + tlen =pci_vpd_lrdt_size (tag) + +PCI_VPD_LRDT_TAG_SIZE ; + }else { + tlen =pci_vpd_srdt_size (tag) + +PCI_VPD_SRDT_TAG_SIZE ; + tag[0 ] &= ~PCI_VPD_SRDT_LEN_MASK ; + } +/* Verify VPD tag fits in area */ +if (tlen + off > dev->vpd->len) +return -ENOENT ; + *len = tlen; +return tag[0 ]; +} + +static int pci_load_vpdr (struct pci_dev *dev) +{ +int rlen, ilen, tag, rc; + +/* Check for VPD-I and VPD-R tag */ + tag =pci_get_vpd_tag (dev,0 , &ilen); +if (tag !=PCI_VPD_LRDT_ID_STRING ) +return -ENOENT ; + tag =pci_get_vpd_tag (dev, ilen, &rlen); +if (tag !=PCI_VPD_LRDT_RO_DATA ) +return -ENOENT ; + + rlen -=PCI_VPD_LRDT_TAG_SIZE ; + dev->vpdr_len = rlen;
Quoted 9 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1488
Quoted 4 lines
drivers/pci/pci-sysfs.c +4
-0+ dev->vpdr_data =kzalloc (rlen,GFP_ATOMIC ); +if (dev->vpdr_data ==NULL ) +return -ENOMEM ; +Why not cache the ID string as well? Seems like it might be a field people would want to read on a regular basis in order to find out what is there.
I could add that. What should attribute be called, 'vpdinfo', 'vpdi', etc?
Quoted 9 lines
Quoted 9 lines
drivers/pci/pci-sysfs.c +9
-0+ rc =pci_read_vpd (dev, ilen +PCI_VPD_LRDT_TAG_SIZE , + rlen, dev->vpdr_data); +if (rc != rlen) + goto error; +if (sysfs_create_group (&dev->dev.kobj, &vpd_attr_group)) + goto error; +return 0 ; + error: +kfree (dev->vpdr_data);
Quoted 8 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1501
Quoted 4 lines
drivers/pci/pci-sysfs.c +4
-0+ dev->vpdr_len =0 ; +return -ENOENT ; +} +This bit here needs to reset
vpdr_databack to NULL. Otherwise you could cause memory corruption via a double free in your two clean-up routines called out below.
Ok will fix that.
Re: [PATCH 2/2] pci: Update VPD size with correct length
Quoted 22 lines · On Tue, Dec 29, 2015 at 11:01 AM, <[email protected]> wrote:
Quoted 14 lines · On Mon, Dec 28, 2015 at 9:29 PM, <[email protected]> wrote:
Quoted 13 lines
I had posted a patch recently to enable exposing the VPD-R valyes to sysfs. I need access to these to parse into systemd for network naming (biosdevname style names).
The VPD-R is a readonly area contained within the PCI Vital Product data. There are some standard and vendor-specific keys stored in this region.
PN = Part Number
SN = Serial Number
MN = Manufacturer ID
Vx = Vendor-specific (x=0..9 A..Z)Quoted 7 lines
Quoted 2 lines
Biosdevname/Systemd will use these VPD keys for determining Network partitioning and port numbers for NIC cards
Can you please repost this as a patch instead of as a reply to our thread about VPD size. The fact is the subject is misleading as your patch isn't actually related to VPD sizing.
Quoted 4 lines
I had already posted this a few weeks back but never got any feedback.
[PATCH] Create sysfs entries for VPD-R keys
https://marc.info/?l=linux-pci&m=144959803316031&w=2
Next time I would recommend resubmitting with the people who might be interested in the Cc instead of just throwing it into an existing thread as it really just muddies the waters for the existing thread to have it branch off like this.
Quoted 39 lines
Quoted 29 lines
Quoted 29 lines
Signed-off-by Jordan Hargrave <[email protected]>--- drivers/pci/pci-sysfs.c | 216 +++++++++++++++++++++++++++++++++++++++++++++++ include/linux/pci.h | 2 + 2 files changed, 218 insertions(+), 0 deletions(-)drivers/pci/pci-sysfs.c +15
-0@@ -1295,6 +1295,210 @@ static struct bin_attribute pcie_config_attr = { .write = pci_write_config, }; +static umode_t vpd_attr_exist (struct kobject *kobj, +struct attribute *attr,int n) +{ +struct device *dev; +struct pci_dev *pdev; +const char *name; +int i; + + dev = container_of(kobj, struct device, kobj); + pdev =to_pci_dev (dev); + + name = attr->name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i = pci_vpd_find_info_keyword(pdev->vpdr_data, 0,Quoted 9 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1316
Quoted 4 lines
drivers/pci/pci-sysfs.c +4
-0+ pdev->vpdr_len, name); +return (i >=0 ?S_IRUGO :0 ); +} +So I assume there is another patch that implements
pci_vpd_find_info_keywordso that it can go through thevpdr_dataand parse it?
Quoted 1 line
That's already an existing function in
drivers/pci/vpd.c
I see that now. Just had to do a bit of grepping through the code. Looks like previously it was used by mostly the Broadcom network device drivers.
Quoted 187 lines
Quoted 128 lines
Quoted 128 lines
drivers/pci/pci-sysfs.c +128
-0+static ssize_t vpd_attr_show (struct device *dev, +struct device_attribute *attr,char *buf) +{ +struct pci_dev *pdev; +const char *name; +char kv_data[257 ] = {0 }; +int i, len; + + pdev =to_pci_dev (dev); + name = attr->attr.name; +if (pdev->vpdr_data ==NULL ) +return 0 ; + i =pci_vpd_find_info_keyword (pdev->vpdr_data,0 , + pdev->vpdr_len, name); +if (i >=0 ) { + len =pci_vpd_info_field_size (&pdev->vpdr_data[i]); +memcpy (kv_data, pdev->vpdr_data + i + +PCI_VPD_INFO_FLD_HDR_SIZE , len); +return scnprintf (buf,PAGE_SIZE ,"%s\n" , kv_data); + } +return 0 ; +} + +#define VPD_ATTR_RO (x) \ +struct device_attribute vpd ## x = { \ + .attr = { .name = #x, .mode = S_IRUGO | S_IWUSR }, \ + .show = vpd_attr_show \ +} + +VPD_ATTR_RO (PN ); +VPD_ATTR_RO (EC ); +VPD_ATTR_RO (MN ); +VPD_ATTR_RO (SN ); +VPD_ATTR_RO (V0 ); +VPD_ATTR_RO (V1 ); +VPD_ATTR_RO (V2 ); +VPD_ATTR_RO (V3 ); +VPD_ATTR_RO (V4 ); +VPD_ATTR_RO (V5 ); +VPD_ATTR_RO (V6 ); +VPD_ATTR_RO (V7 ); +VPD_ATTR_RO (V8 ); +VPD_ATTR_RO (V9 ); +VPD_ATTR_RO (VA ); +VPD_ATTR_RO (VB ); +VPD_ATTR_RO (VC ); +VPD_ATTR_RO (VD ); +VPD_ATTR_RO (VE ); +VPD_ATTR_RO (VF ); +VPD_ATTR_RO (VG ); +VPD_ATTR_RO (VH ); +VPD_ATTR_RO (VI ); +VPD_ATTR_RO (VJ ); +VPD_ATTR_RO (VK ); +VPD_ATTR_RO (VL ); +VPD_ATTR_RO (VM ); +VPD_ATTR_RO (VN ); +VPD_ATTR_RO (VO ); +VPD_ATTR_RO (VP ); +VPD_ATTR_RO (VQ ); +VPD_ATTR_RO (VR ); +VPD_ATTR_RO (VS ); +VPD_ATTR_RO (VT ); +VPD_ATTR_RO (VU ); +VPD_ATTR_RO (VV ); +VPD_ATTR_RO (VW ); +VPD_ATTR_RO (VX ); +VPD_ATTR_RO (VY ); +VPD_ATTR_RO (VZ ); + +static struct attribute *vpd_attributes[] = { + &vpdPN.attr, + &vpdEC.attr, + &vpdMN.attr, + &vpdSN.attr, + &vpdV0.attr, + &vpdV1.attr, + &vpdV2.attr, + &vpdV3.attr, + &vpdV4.attr, + &vpdV5.attr, + &vpdV6.attr, + &vpdV7.attr, + &vpdV8.attr, + &vpdV9.attr, + &vpdVA.attr, + &vpdVB.attr, + &vpdVC.attr, + &vpdVD.attr, + &vpdVE.attr, + &vpdVF.attr, + &vpdVG.attr, + &vpdVH.attr, + &vpdVI.attr, + &vpdVJ.attr, + &vpdVK.attr, + &vpdVL.attr, + &vpdVM.attr, + &vpdVN.attr, + &vpdVO.attr, + &vpdVP.attr, + &vpdVQ.attr, + &vpdVR.attr, + &vpdVS.attr, + &vpdVT.attr, + &vpdVU.attr, + &vpdVV.attr, + &vpdVW.attr, + &vpdVX.attr, + &vpdVY.attr, + &vpdVZ.attr, +NULL , +}; + +static struct attribute_group vpd_attr_group = { + .name ="vpdr" , + .attrs = vpd_attributes, + .is_visible = vpd_attr_exist, +}; + + +static int pci_get_vpd_tag (struct pci_dev *dev,int off,int *len) +{ +u8 tag[3 ]; +int rc, tlen; + + *len =0 ; +/* Quirk Atheros cards, reading VPD hangs system for 20s */ Quoted 9 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1447
Quoted 3 lines
drivers/pci/pci-sysfs.c +3
-0+if (dev->vendor ==PCI_VENDOR_ID_ATHEROS || + dev->vendor ==PCI_VENDOR_ID_ATTANSIC ) +return -ENOENT ;I'm not really sure this is the right place for this type of quirk. If this is really an issue maybe we should just disable VPD for these devices. Otherwise there isn't anything to stop someone from going in and reading the VPD region via the existing VPD interfaces.
This could be moved elsewhere. There's a similar quirk for megaraid cards that was posted recently. [PATCH v4] pci: Limit VPD length for
megaraid_sasadapterQuoted 37 lines
Quoted 37 lines
drivers/pci/pci-sysfs.c +37
-0+ rc =pci_read_vpd (dev, off,1 , tag); +if (rc !=1 ) +return -ENOENT ; +if (tag[0 ] ==0x00 || tag[0 ] ==0xFF || tag[0 ] ==0x7F ) +return -ENOENT ; +if (tag[0 ] &PCI_VPD_LRDT ) { + rc =pci_read_vpd (dev, off+1 ,2 , tag+1 ); +if (rc !=2 ) +return -ENOENT ; + tlen =pci_vpd_lrdt_size (tag) + +PCI_VPD_LRDT_TAG_SIZE ; + }else { + tlen =pci_vpd_srdt_size (tag) + +PCI_VPD_SRDT_TAG_SIZE ; + tag[0 ] &= ~PCI_VPD_SRDT_LEN_MASK ; + } +/* Verify VPD tag fits in area */ +if (tlen + off > dev->vpd->len) +return -ENOENT ; + *len = tlen; +return tag[0 ]; +} + +static int pci_load_vpdr (struct pci_dev *dev) +{ +int rlen, ilen, tag, rc; + +/* Check for VPD-I and VPD-R tag */ + tag =pci_get_vpd_tag (dev,0 , &ilen); +if (tag !=PCI_VPD_LRDT_ID_STRING ) +return -ENOENT ; + tag =pci_get_vpd_tag (dev, ilen, &rlen); +if (tag !=PCI_VPD_LRDT_RO_DATA ) +return -ENOENT ; + + rlen -=PCI_VPD_LRDT_TAG_SIZE ; + dev->vpdr_len = rlen;Quoted 9 lines
drivers/pci/pci-sysfs.c static struct bin_attribute pcie_config_attr = { · line 1488
Quoted 4 lines
drivers/pci/pci-sysfs.c +4
-0+ dev->vpdr_data =kzalloc (rlen,GFP_ATOMIC ); +if (dev->vpdr_data ==NULL ) +return -ENOMEM ; +Why not cache the ID string as well? Seems like it might be a field people would want to read on a regular basis in order to find out what is there.
Quoted 1 line
I could add that. What should attribute be called, 'vpdinfo', 'vpdi', etc?
I'd say keep it simple. Maybe something like 'id-string' since that is the name of the tag.