PCI: Speed up device init by parsing capabilities all at once
Quoted 121 lines · Vikash Bansal
In the current implementation, the PCI capability list is parsed from the beginning to find each capability, which results in a large number of redundant PCI reads.
Instead, we can parse the complete list just once, store it in the
pci_devstructure, and get the offset of each capability directly from thepci_devstructure.This implementation improves pci devices initialization time by ~2-3% (from 270ms to 261ms) in case of bare metal and 7-8% (From 201ms to 184ms) in case of VM running on ESXi.
It also adds a memory overhead of 20bytes (value of
PCI_CAP_ID_MAX) per PCI device.Run pahole for
pci_devstructure. This patch is not adding any padding bytes.Signed-off-by Vikash Bansal <[email protected]>--- drivers/pci/pci.c | 43 ++++++++++++++++++++++++++++++++++++------- drivers/pci/probe.c | 1 + include/linux/pci.h | 2 ++ 3 files changed, 39 insertions(+), 7 deletions(-)drivers/pci/pci.c +36
-7@@ -468,6 +468,41 @@ static u8 __pci_bus_find_cap_start(struct pci_bus *bus,return 0 ; } + +/** +* pci_find_all_capabilities - Read all capabilities +* @dev: the PCI device +* +* Read all capabilities and store offsets in cap_off +* array in pci_dev structure. +*/ +void pci_find_all_capabilities (struct pci_dev *dev) +{ +int ttl =PCI_FIND_CAP_TTL ; +u16 ent; +u8 pos; +u8 id; + + pos =__pci_bus_find_cap_start (dev->bus, dev->devfn, dev->hdr_type); +if (!pos) +return ; +pci_bus_read_config_byte (dev->bus, dev->devfn, pos, &pos); +while (ttl--) { +if (pos <0x40 ) +break ; + pos &= ~3 ; +pci_bus_read_config_word (dev->bus, dev->devfn, pos, &ent); + id = ent &0xff ; +if (id ==0xff ) +break ; + +/* Read first instance of capability */ +if (!(dev->cap_off[id])) + dev->cap_off[id] = pos; + pos = (ent >>8 ); + } +} +/** * pci_find_capability - query for devices' capabilities * @dev: PCI device to query @@ -489,13 +524,7 @@ static u8 __pci_bus_find_cap_start(struct pci_bus *bus,*/ u8 pci_find_capability (struct pci_dev *dev,int cap) { -u8 pos; - - pos =__pci_bus_find_cap_start (dev->bus, dev->devfn, dev->hdr_type); -if (pos) - pos =__pci_find_next_cap (dev->bus, dev->devfn, pos, cap); - -return pos; +return dev->cap_off[cap]; }EXPORT_SYMBOL (pci_find_capability);drivers/pci/probe.c +1
-0@@ -1831,6 +1831,7 @@ int pci_setup_device(struct pci_dev *dev) dev->hdr_type = hdr_type &0x7f ; dev->multifunction = !!(hdr_type &0x80 ); dev->error_state = pci_channel_io_normal; +pci_find_all_capabilities (dev);set_pcie_port_type (dev);pci_set_of_node (dev);include/linux/pci.h +2
-0@@ -335,6 +335,7 @@ struct pci_dev {unsigned int class;/* 3 bytes: (base,sub,prog-if) */ u8 revision;/* PCI revision, low byte of class word */ u8 hdr_type;/* PCI header type (`multi' flag masked out) */ +u8 cap_off[PCI_CAP_ID_MAX ];/* Offsets of all pci capabilities */ #ifdef CONFIG_PCIEAER u16 aer_cap;/* AER capability offset */ struct aer_stats *aer_stats;/* AER stats for this device */ @@ -1140,6 +1141,7 @@ void pci_sort_breadthfirst(void);u8 pci_bus_find_capability (struct pci_bus *bus,unsigned int devfn,int cap);u8 pci_find_capability (struct pci_dev *dev,int cap); +void pci_find_all_capabilities (struct pci_dev *dev);u8 pci_find_next_capability (struct pci_dev *dev,u8 pos,int cap);u8 pci_find_ht_capability (struct pci_dev *dev,int ht_cap);u8 pci_find_next_ht_capability (struct pci_dev *dev,u8 pos,int ht_cap);Signature · 1 line
2.30.0
Hi,
This is the friendly patch-bot of Greg Kroah-Hartman. You have sent him a patch that has triggered this response. He used to manually respond to these common problems, but in order to save his sanity (he kept writing the same thing over and over, yet to different people), I was created. Hopefully you will not take offence and will fix the problem in your patch and resubmit it so that it can be accepted into the Linux kernel tree.
You are receiving this message because of the following common error(s) as indicated below:
- This looks like a new version of a previously submitted patch, but you did not list below the --- line any changes from the previous version. Please read the section entitled "The canonical patch format" in the kernel file, Documentation/SubmittingPatches for what needs to be done here to properly describe this.
If you wish to discuss this problem further, or you have questions about how to resolve this issue, please feel free to respond to this email and Greg will reply once he has dug out from the pending patches received from other developers.
thanks,
greg k-h's patch email bot