January 2022 - Linux-stable-mirror

FAILED: patch "[PATCH] net/fsl: xgmac_mdio: Add workaround for erratum A-009885" failed to apply to 4.14-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.14-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 6198c722019774d38018457a8bfb9ba3ed8c931e Mon Sep 17 00:00:00 2001 From: Tobias Waldekranz <tobias(a)waldekranz.com> Date: Tue, 18 Jan 2022 22:50:50 +0100 Subject: [PATCH] net/fsl: xgmac_mdio: Add workaround for erratum A-009885 Once an MDIO read transaction is initiated, we must read back the data register within 16 MDC cycles after the transaction completes. Outside of this window, reads may return corrupt data. Therefore, disable local interrupts in the critical section, to maximize the probability that we can satisfy this requirement. Fixes: d55ad2967d89 ("powerpc/mpc85xx: Create dts components for the FSL QorIQ DPAA FMan") Signed-off-by: Tobias Waldekranz <tobias(a)waldekranz.com> Reviewed-by: Andrew Lunn <andrew(a)lunn.ch> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/drivers/net/ethernet/freescale/xgmac_mdio.c b/drivers/net/ethernet/freescale/xgmac_mdio.c index 5b8b9bcf41a2..bf566ac3195b 100644 --- a/drivers/net/ethernet/freescale/xgmac_mdio.c +++ b/drivers/net/ethernet/freescale/xgmac_mdio.c @@ -51,6 +51,7 @@ struct tgec_mdio_controller { struct mdio_fsl_priv { struct tgec_mdio_controller __iomem *mdio_base; bool is_little_endian; + bool has_a009885; bool has_a011043; }; @@ -186,10 +187,10 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) { struct mdio_fsl_priv *priv = (struct mdio_fsl_priv *)bus->priv; struct tgec_mdio_controller __iomem *regs = priv->mdio_base; + unsigned long flags; uint16_t dev_addr; uint32_t mdio_stat; uint32_t mdio_ctl; - uint16_t value; int ret; bool endian = priv->is_little_endian; @@ -221,12 +222,18 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) return ret; } + if (priv->has_a009885) + /* Once the operation completes, i.e. MDIO_STAT_BSY clears, we + * must read back the data register within 16 MDC cycles. + */ + local_irq_save(flags); + /* Initiate the read */ xgmac_write32(mdio_ctl | MDIO_CTL_READ, &regs->mdio_ctl, endian); ret = xgmac_wait_until_done(&bus->dev, regs, endian); if (ret) - return ret; + goto irq_restore; /* Return all Fs if nothing was there */ if ((xgmac_read32(&regs->mdio_stat, endian) & MDIO_STAT_RD_ER) && @@ -234,13 +241,17 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) dev_dbg(&bus->dev, "Error while reading PHY%d reg at %d.%hhu\n", phy_id, dev_addr, regnum); - return 0xffff; + ret = 0xffff; + } else { + ret = xgmac_read32(&regs->mdio_data, endian) & 0xffff; + dev_dbg(&bus->dev, "read %04x\n", ret); } - value = xgmac_read32(&regs->mdio_data, endian) & 0xffff; - dev_dbg(&bus->dev, "read %04x\n", value); +irq_restore: + if (priv->has_a009885) + local_irq_restore(flags); - return value; + return ret; } static int xgmac_mdio_probe(struct platform_device *pdev) @@ -287,6 +298,8 @@ static int xgmac_mdio_probe(struct platform_device *pdev) priv->is_little_endian = device_property_read_bool(&pdev->dev, "little-endian"); + priv->has_a009885 = device_property_read_bool(&pdev->dev, + "fsl,erratum-a009885"); priv->has_a011043 = device_property_read_bool(&pdev->dev, "fsl,erratum-a011043");

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] net/fsl: xgmac_mdio: Add workaround for erratum A-009885" failed to apply to 4.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 6198c722019774d38018457a8bfb9ba3ed8c931e Mon Sep 17 00:00:00 2001 From: Tobias Waldekranz <tobias(a)waldekranz.com> Date: Tue, 18 Jan 2022 22:50:50 +0100 Subject: [PATCH] net/fsl: xgmac_mdio: Add workaround for erratum A-009885 Once an MDIO read transaction is initiated, we must read back the data register within 16 MDC cycles after the transaction completes. Outside of this window, reads may return corrupt data. Therefore, disable local interrupts in the critical section, to maximize the probability that we can satisfy this requirement. Fixes: d55ad2967d89 ("powerpc/mpc85xx: Create dts components for the FSL QorIQ DPAA FMan") Signed-off-by: Tobias Waldekranz <tobias(a)waldekranz.com> Reviewed-by: Andrew Lunn <andrew(a)lunn.ch> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/drivers/net/ethernet/freescale/xgmac_mdio.c b/drivers/net/ethernet/freescale/xgmac_mdio.c index 5b8b9bcf41a2..bf566ac3195b 100644 --- a/drivers/net/ethernet/freescale/xgmac_mdio.c +++ b/drivers/net/ethernet/freescale/xgmac_mdio.c @@ -51,6 +51,7 @@ struct tgec_mdio_controller { struct mdio_fsl_priv { struct tgec_mdio_controller __iomem *mdio_base; bool is_little_endian; + bool has_a009885; bool has_a011043; }; @@ -186,10 +187,10 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) { struct mdio_fsl_priv *priv = (struct mdio_fsl_priv *)bus->priv; struct tgec_mdio_controller __iomem *regs = priv->mdio_base; + unsigned long flags; uint16_t dev_addr; uint32_t mdio_stat; uint32_t mdio_ctl; - uint16_t value; int ret; bool endian = priv->is_little_endian; @@ -221,12 +222,18 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) return ret; } + if (priv->has_a009885) + /* Once the operation completes, i.e. MDIO_STAT_BSY clears, we + * must read back the data register within 16 MDC cycles. + */ + local_irq_save(flags); + /* Initiate the read */ xgmac_write32(mdio_ctl | MDIO_CTL_READ, &regs->mdio_ctl, endian); ret = xgmac_wait_until_done(&bus->dev, regs, endian); if (ret) - return ret; + goto irq_restore; /* Return all Fs if nothing was there */ if ((xgmac_read32(&regs->mdio_stat, endian) & MDIO_STAT_RD_ER) && @@ -234,13 +241,17 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) dev_dbg(&bus->dev, "Error while reading PHY%d reg at %d.%hhu\n", phy_id, dev_addr, regnum); - return 0xffff; + ret = 0xffff; + } else { + ret = xgmac_read32(&regs->mdio_data, endian) & 0xffff; + dev_dbg(&bus->dev, "read %04x\n", ret); } - value = xgmac_read32(&regs->mdio_data, endian) & 0xffff; - dev_dbg(&bus->dev, "read %04x\n", value); +irq_restore: + if (priv->has_a009885) + local_irq_restore(flags); - return value; + return ret; } static int xgmac_mdio_probe(struct platform_device *pdev) @@ -287,6 +298,8 @@ static int xgmac_mdio_probe(struct platform_device *pdev) priv->is_little_endian = device_property_read_bool(&pdev->dev, "little-endian"); + priv->has_a009885 = device_property_read_bool(&pdev->dev, + "fsl,erratum-a009885"); priv->has_a011043 = device_property_read_bool(&pdev->dev, "fsl,erratum-a011043");

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] net/fsl: xgmac_mdio: Add workaround for erratum A-009885" failed to apply to 4.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 6198c722019774d38018457a8bfb9ba3ed8c931e Mon Sep 17 00:00:00 2001 From: Tobias Waldekranz <tobias(a)waldekranz.com> Date: Tue, 18 Jan 2022 22:50:50 +0100 Subject: [PATCH] net/fsl: xgmac_mdio: Add workaround for erratum A-009885 Once an MDIO read transaction is initiated, we must read back the data register within 16 MDC cycles after the transaction completes. Outside of this window, reads may return corrupt data. Therefore, disable local interrupts in the critical section, to maximize the probability that we can satisfy this requirement. Fixes: d55ad2967d89 ("powerpc/mpc85xx: Create dts components for the FSL QorIQ DPAA FMan") Signed-off-by: Tobias Waldekranz <tobias(a)waldekranz.com> Reviewed-by: Andrew Lunn <andrew(a)lunn.ch> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/drivers/net/ethernet/freescale/xgmac_mdio.c b/drivers/net/ethernet/freescale/xgmac_mdio.c index 5b8b9bcf41a2..bf566ac3195b 100644 --- a/drivers/net/ethernet/freescale/xgmac_mdio.c +++ b/drivers/net/ethernet/freescale/xgmac_mdio.c @@ -51,6 +51,7 @@ struct tgec_mdio_controller { struct mdio_fsl_priv { struct tgec_mdio_controller __iomem *mdio_base; bool is_little_endian; + bool has_a009885; bool has_a011043; }; @@ -186,10 +187,10 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) { struct mdio_fsl_priv *priv = (struct mdio_fsl_priv *)bus->priv; struct tgec_mdio_controller __iomem *regs = priv->mdio_base; + unsigned long flags; uint16_t dev_addr; uint32_t mdio_stat; uint32_t mdio_ctl; - uint16_t value; int ret; bool endian = priv->is_little_endian; @@ -221,12 +222,18 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) return ret; } + if (priv->has_a009885) + /* Once the operation completes, i.e. MDIO_STAT_BSY clears, we + * must read back the data register within 16 MDC cycles. + */ + local_irq_save(flags); + /* Initiate the read */ xgmac_write32(mdio_ctl | MDIO_CTL_READ, &regs->mdio_ctl, endian); ret = xgmac_wait_until_done(&bus->dev, regs, endian); if (ret) - return ret; + goto irq_restore; /* Return all Fs if nothing was there */ if ((xgmac_read32(&regs->mdio_stat, endian) & MDIO_STAT_RD_ER) && @@ -234,13 +241,17 @@ static int xgmac_mdio_read(struct mii_bus *bus, int phy_id, int regnum) dev_dbg(&bus->dev, "Error while reading PHY%d reg at %d.%hhu\n", phy_id, dev_addr, regnum); - return 0xffff; + ret = 0xffff; + } else { + ret = xgmac_read32(&regs->mdio_data, endian) & 0xffff; + dev_dbg(&bus->dev, "read %04x\n", ret); } - value = xgmac_read32(&regs->mdio_data, endian) & 0xffff; - dev_dbg(&bus->dev, "read %04x\n", value); +irq_restore: + if (priv->has_a009885) + local_irq_restore(flags); - return value; + return ret; } static int xgmac_mdio_probe(struct platform_device *pdev) @@ -287,6 +298,8 @@ static int xgmac_mdio_probe(struct platform_device *pdev) priv->is_little_endian = device_property_read_bool(&pdev->dev, "little-endian"); + priv->has_a009885 = device_property_read_bool(&pdev->dev, + "fsl,erratum-a009885"); priv->has_a011043 = device_property_read_bool(&pdev->dev, "fsl,erratum-a011043");

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] ipv4: avoid quadratic behavior in netns dismantle" failed to apply to 4.14-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.14-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From d07418afea8f1d9896aaf9dc5ae47ac4f45b220c Mon Sep 17 00:00:00 2001 From: Eric Dumazet <edumazet(a)google.com> Date: Wed, 19 Jan 2022 02:04:12 -0800 Subject: [PATCH] ipv4: avoid quadratic behavior in netns dismantle net/ipv4/fib_semantics.c uses an hash table of 256 slots, keyed by device ifindexes: fib_info_devhash[DEVINDEX_HASHSIZE] Problem is that with network namespaces, devices tend to use the same ifindex. lo device for instance has a fixed ifindex of one, for all network namespaces. This means that hosts with thousands of netns spend a lot of time looking at some hash buckets with thousands of elements, notably at netns dismantle. Simply add a per netns perturbation (net_hash_mix()) to spread elements more uniformely. Also change fib_devindex_hashfn() to use more entropy. Fixes: aa79e66eee5d ("net: Make ifindex generation per-net namespace") Signed-off-by: Eric Dumazet <edumazet(a)google.com> Reviewed-by: David Ahern <dsahern(a)kernel.org> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c index 45619c005b8d..9813949da104 100644 --- a/net/ipv4/fib_semantics.c +++ b/net/ipv4/fib_semantics.c @@ -29,6 +29,7 @@ #include <linux/init.h> #include <linux/slab.h> #include <linux/netlink.h> +#include <linux/hash.h> #include <net/arp.h> #include <net/ip.h> @@ -319,11 +320,15 @@ static inline int nh_comp(struct fib_info *fi, struct fib_info *ofi) static inline unsigned int fib_devindex_hashfn(unsigned int val) { - unsigned int mask = DEVINDEX_HASHSIZE - 1; + return hash_32(val, DEVINDEX_HASHBITS); +} + +static struct hlist_head * +fib_info_devhash_bucket(const struct net_device *dev) +{ + u32 val = net_hash_mix(dev_net(dev)) ^ dev->ifindex; - return (val ^ - (val >> DEVINDEX_HASHBITS) ^ - (val >> (DEVINDEX_HASHBITS * 2))) & mask; + return &fib_info_devhash[fib_devindex_hashfn(val)]; } static unsigned int fib_info_hashfn_1(int init_val, u8 protocol, u8 scope, @@ -433,12 +438,11 @@ int ip_fib_check_default(__be32 gw, struct net_device *dev) { struct hlist_head *head; struct fib_nh *nh; - unsigned int hash; spin_lock(&fib_info_lock); - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); + hlist_for_each_entry(nh, head, nh_hash) { if (nh->fib_nh_dev == dev && nh->fib_nh_gw4 == gw && @@ -1609,12 +1613,10 @@ struct fib_info *fib_create_info(struct fib_config *cfg, } else { change_nexthops(fi) { struct hlist_head *head; - unsigned int hash; if (!nexthop_nh->fib_nh_dev) continue; - hash = fib_devindex_hashfn(nexthop_nh->fib_nh_dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(nexthop_nh->fib_nh_dev); hlist_add_head(&nexthop_nh->nh_hash, head); } endfor_nexthops(fi) } @@ -1966,8 +1968,7 @@ void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig) void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) { - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_nh *nh; hlist_for_each_entry(nh, head, nh_hash) { @@ -1986,12 +1987,11 @@ void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) */ int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force) { - int ret = 0; - int scope = RT_SCOPE_NOWHERE; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_info *prev_fi = NULL; - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + int scope = RT_SCOPE_NOWHERE; struct fib_nh *nh; + int ret = 0; if (force) scope = -1; @@ -2136,7 +2136,6 @@ static void fib_select_default(const struct flowi4 *flp, struct fib_result *res) int fib_sync_up(struct net_device *dev, unsigned char nh_flags) { struct fib_info *prev_fi; - unsigned int hash; struct hlist_head *head; struct fib_nh *nh; int ret; @@ -2152,8 +2151,7 @@ int fib_sync_up(struct net_device *dev, unsigned char nh_flags) } prev_fi = NULL; - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); ret = 0; hlist_for_each_entry(nh, head, nh_hash) {

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] ipv4: avoid quadratic behavior in netns dismantle" failed to apply to 4.19-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.19-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From d07418afea8f1d9896aaf9dc5ae47ac4f45b220c Mon Sep 17 00:00:00 2001 From: Eric Dumazet <edumazet(a)google.com> Date: Wed, 19 Jan 2022 02:04:12 -0800 Subject: [PATCH] ipv4: avoid quadratic behavior in netns dismantle net/ipv4/fib_semantics.c uses an hash table of 256 slots, keyed by device ifindexes: fib_info_devhash[DEVINDEX_HASHSIZE] Problem is that with network namespaces, devices tend to use the same ifindex. lo device for instance has a fixed ifindex of one, for all network namespaces. This means that hosts with thousands of netns spend a lot of time looking at some hash buckets with thousands of elements, notably at netns dismantle. Simply add a per netns perturbation (net_hash_mix()) to spread elements more uniformely. Also change fib_devindex_hashfn() to use more entropy. Fixes: aa79e66eee5d ("net: Make ifindex generation per-net namespace") Signed-off-by: Eric Dumazet <edumazet(a)google.com> Reviewed-by: David Ahern <dsahern(a)kernel.org> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c index 45619c005b8d..9813949da104 100644 --- a/net/ipv4/fib_semantics.c +++ b/net/ipv4/fib_semantics.c @@ -29,6 +29,7 @@ #include <linux/init.h> #include <linux/slab.h> #include <linux/netlink.h> +#include <linux/hash.h> #include <net/arp.h> #include <net/ip.h> @@ -319,11 +320,15 @@ static inline int nh_comp(struct fib_info *fi, struct fib_info *ofi) static inline unsigned int fib_devindex_hashfn(unsigned int val) { - unsigned int mask = DEVINDEX_HASHSIZE - 1; + return hash_32(val, DEVINDEX_HASHBITS); +} + +static struct hlist_head * +fib_info_devhash_bucket(const struct net_device *dev) +{ + u32 val = net_hash_mix(dev_net(dev)) ^ dev->ifindex; - return (val ^ - (val >> DEVINDEX_HASHBITS) ^ - (val >> (DEVINDEX_HASHBITS * 2))) & mask; + return &fib_info_devhash[fib_devindex_hashfn(val)]; } static unsigned int fib_info_hashfn_1(int init_val, u8 protocol, u8 scope, @@ -433,12 +438,11 @@ int ip_fib_check_default(__be32 gw, struct net_device *dev) { struct hlist_head *head; struct fib_nh *nh; - unsigned int hash; spin_lock(&fib_info_lock); - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); + hlist_for_each_entry(nh, head, nh_hash) { if (nh->fib_nh_dev == dev && nh->fib_nh_gw4 == gw && @@ -1609,12 +1613,10 @@ struct fib_info *fib_create_info(struct fib_config *cfg, } else { change_nexthops(fi) { struct hlist_head *head; - unsigned int hash; if (!nexthop_nh->fib_nh_dev) continue; - hash = fib_devindex_hashfn(nexthop_nh->fib_nh_dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(nexthop_nh->fib_nh_dev); hlist_add_head(&nexthop_nh->nh_hash, head); } endfor_nexthops(fi) } @@ -1966,8 +1968,7 @@ void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig) void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) { - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_nh *nh; hlist_for_each_entry(nh, head, nh_hash) { @@ -1986,12 +1987,11 @@ void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) */ int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force) { - int ret = 0; - int scope = RT_SCOPE_NOWHERE; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_info *prev_fi = NULL; - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + int scope = RT_SCOPE_NOWHERE; struct fib_nh *nh; + int ret = 0; if (force) scope = -1; @@ -2136,7 +2136,6 @@ static void fib_select_default(const struct flowi4 *flp, struct fib_result *res) int fib_sync_up(struct net_device *dev, unsigned char nh_flags) { struct fib_info *prev_fi; - unsigned int hash; struct hlist_head *head; struct fib_nh *nh; int ret; @@ -2152,8 +2151,7 @@ int fib_sync_up(struct net_device *dev, unsigned char nh_flags) } prev_fi = NULL; - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); ret = 0; hlist_for_each_entry(nh, head, nh_hash) {

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] ipv4: avoid quadratic behavior in netns dismantle" failed to apply to 4.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From d07418afea8f1d9896aaf9dc5ae47ac4f45b220c Mon Sep 17 00:00:00 2001 From: Eric Dumazet <edumazet(a)google.com> Date: Wed, 19 Jan 2022 02:04:12 -0800 Subject: [PATCH] ipv4: avoid quadratic behavior in netns dismantle net/ipv4/fib_semantics.c uses an hash table of 256 slots, keyed by device ifindexes: fib_info_devhash[DEVINDEX_HASHSIZE] Problem is that with network namespaces, devices tend to use the same ifindex. lo device for instance has a fixed ifindex of one, for all network namespaces. This means that hosts with thousands of netns spend a lot of time looking at some hash buckets with thousands of elements, notably at netns dismantle. Simply add a per netns perturbation (net_hash_mix()) to spread elements more uniformely. Also change fib_devindex_hashfn() to use more entropy. Fixes: aa79e66eee5d ("net: Make ifindex generation per-net namespace") Signed-off-by: Eric Dumazet <edumazet(a)google.com> Reviewed-by: David Ahern <dsahern(a)kernel.org> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c index 45619c005b8d..9813949da104 100644 --- a/net/ipv4/fib_semantics.c +++ b/net/ipv4/fib_semantics.c @@ -29,6 +29,7 @@ #include <linux/init.h> #include <linux/slab.h> #include <linux/netlink.h> +#include <linux/hash.h> #include <net/arp.h> #include <net/ip.h> @@ -319,11 +320,15 @@ static inline int nh_comp(struct fib_info *fi, struct fib_info *ofi) static inline unsigned int fib_devindex_hashfn(unsigned int val) { - unsigned int mask = DEVINDEX_HASHSIZE - 1; + return hash_32(val, DEVINDEX_HASHBITS); +} + +static struct hlist_head * +fib_info_devhash_bucket(const struct net_device *dev) +{ + u32 val = net_hash_mix(dev_net(dev)) ^ dev->ifindex; - return (val ^ - (val >> DEVINDEX_HASHBITS) ^ - (val >> (DEVINDEX_HASHBITS * 2))) & mask; + return &fib_info_devhash[fib_devindex_hashfn(val)]; } static unsigned int fib_info_hashfn_1(int init_val, u8 protocol, u8 scope, @@ -433,12 +438,11 @@ int ip_fib_check_default(__be32 gw, struct net_device *dev) { struct hlist_head *head; struct fib_nh *nh; - unsigned int hash; spin_lock(&fib_info_lock); - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); + hlist_for_each_entry(nh, head, nh_hash) { if (nh->fib_nh_dev == dev && nh->fib_nh_gw4 == gw && @@ -1609,12 +1613,10 @@ struct fib_info *fib_create_info(struct fib_config *cfg, } else { change_nexthops(fi) { struct hlist_head *head; - unsigned int hash; if (!nexthop_nh->fib_nh_dev) continue; - hash = fib_devindex_hashfn(nexthop_nh->fib_nh_dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(nexthop_nh->fib_nh_dev); hlist_add_head(&nexthop_nh->nh_hash, head); } endfor_nexthops(fi) } @@ -1966,8 +1968,7 @@ void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig) void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) { - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_nh *nh; hlist_for_each_entry(nh, head, nh_hash) { @@ -1986,12 +1987,11 @@ void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) */ int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force) { - int ret = 0; - int scope = RT_SCOPE_NOWHERE; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_info *prev_fi = NULL; - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + int scope = RT_SCOPE_NOWHERE; struct fib_nh *nh; + int ret = 0; if (force) scope = -1; @@ -2136,7 +2136,6 @@ static void fib_select_default(const struct flowi4 *flp, struct fib_result *res) int fib_sync_up(struct net_device *dev, unsigned char nh_flags) { struct fib_info *prev_fi; - unsigned int hash; struct hlist_head *head; struct fib_nh *nh; int ret; @@ -2152,8 +2151,7 @@ int fib_sync_up(struct net_device *dev, unsigned char nh_flags) } prev_fi = NULL; - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); ret = 0; hlist_for_each_entry(nh, head, nh_hash) {

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] ipv4: avoid quadratic behavior in netns dismantle" failed to apply to 4.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From d07418afea8f1d9896aaf9dc5ae47ac4f45b220c Mon Sep 17 00:00:00 2001 From: Eric Dumazet <edumazet(a)google.com> Date: Wed, 19 Jan 2022 02:04:12 -0800 Subject: [PATCH] ipv4: avoid quadratic behavior in netns dismantle net/ipv4/fib_semantics.c uses an hash table of 256 slots, keyed by device ifindexes: fib_info_devhash[DEVINDEX_HASHSIZE] Problem is that with network namespaces, devices tend to use the same ifindex. lo device for instance has a fixed ifindex of one, for all network namespaces. This means that hosts with thousands of netns spend a lot of time looking at some hash buckets with thousands of elements, notably at netns dismantle. Simply add a per netns perturbation (net_hash_mix()) to spread elements more uniformely. Also change fib_devindex_hashfn() to use more entropy. Fixes: aa79e66eee5d ("net: Make ifindex generation per-net namespace") Signed-off-by: Eric Dumazet <edumazet(a)google.com> Reviewed-by: David Ahern <dsahern(a)kernel.org> Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c index 45619c005b8d..9813949da104 100644 --- a/net/ipv4/fib_semantics.c +++ b/net/ipv4/fib_semantics.c @@ -29,6 +29,7 @@ #include <linux/init.h> #include <linux/slab.h> #include <linux/netlink.h> +#include <linux/hash.h> #include <net/arp.h> #include <net/ip.h> @@ -319,11 +320,15 @@ static inline int nh_comp(struct fib_info *fi, struct fib_info *ofi) static inline unsigned int fib_devindex_hashfn(unsigned int val) { - unsigned int mask = DEVINDEX_HASHSIZE - 1; + return hash_32(val, DEVINDEX_HASHBITS); +} + +static struct hlist_head * +fib_info_devhash_bucket(const struct net_device *dev) +{ + u32 val = net_hash_mix(dev_net(dev)) ^ dev->ifindex; - return (val ^ - (val >> DEVINDEX_HASHBITS) ^ - (val >> (DEVINDEX_HASHBITS * 2))) & mask; + return &fib_info_devhash[fib_devindex_hashfn(val)]; } static unsigned int fib_info_hashfn_1(int init_val, u8 protocol, u8 scope, @@ -433,12 +438,11 @@ int ip_fib_check_default(__be32 gw, struct net_device *dev) { struct hlist_head *head; struct fib_nh *nh; - unsigned int hash; spin_lock(&fib_info_lock); - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); + hlist_for_each_entry(nh, head, nh_hash) { if (nh->fib_nh_dev == dev && nh->fib_nh_gw4 == gw && @@ -1609,12 +1613,10 @@ struct fib_info *fib_create_info(struct fib_config *cfg, } else { change_nexthops(fi) { struct hlist_head *head; - unsigned int hash; if (!nexthop_nh->fib_nh_dev) continue; - hash = fib_devindex_hashfn(nexthop_nh->fib_nh_dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(nexthop_nh->fib_nh_dev); hlist_add_head(&nexthop_nh->nh_hash, head); } endfor_nexthops(fi) } @@ -1966,8 +1968,7 @@ void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig) void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) { - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_nh *nh; hlist_for_each_entry(nh, head, nh_hash) { @@ -1986,12 +1987,11 @@ void fib_sync_mtu(struct net_device *dev, u32 orig_mtu) */ int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force) { - int ret = 0; - int scope = RT_SCOPE_NOWHERE; + struct hlist_head *head = fib_info_devhash_bucket(dev); struct fib_info *prev_fi = NULL; - unsigned int hash = fib_devindex_hashfn(dev->ifindex); - struct hlist_head *head = &fib_info_devhash[hash]; + int scope = RT_SCOPE_NOWHERE; struct fib_nh *nh; + int ret = 0; if (force) scope = -1; @@ -2136,7 +2136,6 @@ static void fib_select_default(const struct flowi4 *flp, struct fib_result *res) int fib_sync_up(struct net_device *dev, unsigned char nh_flags) { struct fib_info *prev_fi; - unsigned int hash; struct hlist_head *head; struct fib_nh *nh; int ret; @@ -2152,8 +2151,7 @@ int fib_sync_up(struct net_device *dev, unsigned char nh_flags) } prev_fi = NULL; - hash = fib_devindex_hashfn(dev->ifindex); - head = &fib_info_devhash[hash]; + head = fib_info_devhash_bucket(dev); ret = 0; hlist_for_each_entry(nh, head, nh_hash) {

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] bpf: Fix mount source show for bpffs" failed to apply to 5.10-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.10-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 1e9d74660d4df625b0889e77018f9e94727ceacd Mon Sep 17 00:00:00 2001 From: Yafang Shao <laoar.shao(a)gmail.com> Date: Sat, 8 Jan 2022 13:46:23 +0000 Subject: [PATCH] bpf: Fix mount source show for bpffs We noticed our tc ebpf tools can't start after we upgrade our in-house kernel version from 4.19 to 5.10. That is because of the behaviour change in bpffs caused by commit d2935de7e4fd ("vfs: Convert bpf to use the new mount API"). In our tc ebpf tools, we do strict environment check. If the environment is not matched, we won't allow to start the ebpf progs. One of the check is whether bpffs is properly mounted. The mount information of bpffs in kernel-4.19 and kernel-5.10 are as follows: - kernel 4.19 $ mount -t bpf bpffs /sys/fs/bpf $ mount -t bpf bpffs on /sys/fs/bpf type bpf (rw,relatime) - kernel 5.10 $ mount -t bpf bpffs /sys/fs/bpf $ mount -t bpf none on /sys/fs/bpf type bpf (rw,relatime) The device name in kernel-5.10 is displayed as none instead of bpffs, then our environment check fails. Currently we modify the tools to adopt to the kernel behaviour change, but I think we'd better change the kernel code to keep the behavior consistent. After this change, the mount information will be displayed the same with the behavior in kernel-4.19, for example: $ mount -t bpf bpffs /sys/fs/bpf $ mount -t bpf bpffs on /sys/fs/bpf type bpf (rw,relatime) Fixes: d2935de7e4fd ("vfs: Convert bpf to use the new mount API") Suggested-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Yafang Shao <laoar.shao(a)gmail.com> Signed-off-by: Daniel Borkmann <daniel(a)iogearbox.net> Acked-by: Christian Brauner <christian.brauner(a)ubuntu.com> Cc: David Howells <dhowells(a)redhat.com> Cc: Al Viro <viro(a)zeniv.linux.org.uk> Link: https://lore.kernel.org/bpf/20220108134623.32467-1-laoar.shao@gmail.com diff --git a/kernel/bpf/inode.c b/kernel/bpf/inode.c index 80da1db47c68..5a8d9f7467bf 100644 --- a/kernel/bpf/inode.c +++ b/kernel/bpf/inode.c @@ -648,12 +648,22 @@ static int bpf_parse_param(struct fs_context *fc, struct fs_parameter *param) int opt; opt = fs_parse(fc, bpf_fs_parameters, param, &result); - if (opt < 0) + if (opt < 0) { /* We might like to report bad mount options here, but * traditionally we've ignored all mount options, so we'd * better continue to ignore non-existing options for bpf. */ - return opt == -ENOPARAM ? 0 : opt; + if (opt == -ENOPARAM) { + opt = vfs_parse_fs_param_source(fc, param); + if (opt != -ENOPARAM) + return opt; + + return 0; + } + + if (opt < 0) + return opt; + } switch (opt) { case OPT_MODE:

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] bpf: Fix mount source show for bpffs" failed to apply to 5.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 1e9d74660d4df625b0889e77018f9e94727ceacd Mon Sep 17 00:00:00 2001 From: Yafang Shao <laoar.shao(a)gmail.com> Date: Sat, 8 Jan 2022 13:46:23 +0000 Subject: [PATCH] bpf: Fix mount source show for bpffs We noticed our tc ebpf tools can't start after we upgrade our in-house kernel version from 4.19 to 5.10. That is because of the behaviour change in bpffs caused by commit d2935de7e4fd ("vfs: Convert bpf to use the new mount API"). In our tc ebpf tools, we do strict environment check. If the environment is not matched, we won't allow to start the ebpf progs. One of the check is whether bpffs is properly mounted. The mount information of bpffs in kernel-4.19 and kernel-5.10 are as follows: - kernel 4.19 $ mount -t bpf bpffs /sys/fs/bpf $ mount -t bpf bpffs on /sys/fs/bpf type bpf (rw,relatime) - kernel 5.10 $ mount -t bpf bpffs /sys/fs/bpf $ mount -t bpf none on /sys/fs/bpf type bpf (rw,relatime) The device name in kernel-5.10 is displayed as none instead of bpffs, then our environment check fails. Currently we modify the tools to adopt to the kernel behaviour change, but I think we'd better change the kernel code to keep the behavior consistent. After this change, the mount information will be displayed the same with the behavior in kernel-4.19, for example: $ mount -t bpf bpffs /sys/fs/bpf $ mount -t bpf bpffs on /sys/fs/bpf type bpf (rw,relatime) Fixes: d2935de7e4fd ("vfs: Convert bpf to use the new mount API") Suggested-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Yafang Shao <laoar.shao(a)gmail.com> Signed-off-by: Daniel Borkmann <daniel(a)iogearbox.net> Acked-by: Christian Brauner <christian.brauner(a)ubuntu.com> Cc: David Howells <dhowells(a)redhat.com> Cc: Al Viro <viro(a)zeniv.linux.org.uk> Link: https://lore.kernel.org/bpf/20220108134623.32467-1-laoar.shao@gmail.com diff --git a/kernel/bpf/inode.c b/kernel/bpf/inode.c index 80da1db47c68..5a8d9f7467bf 100644 --- a/kernel/bpf/inode.c +++ b/kernel/bpf/inode.c @@ -648,12 +648,22 @@ static int bpf_parse_param(struct fs_context *fc, struct fs_parameter *param) int opt; opt = fs_parse(fc, bpf_fs_parameters, param, &result); - if (opt < 0) + if (opt < 0) { /* We might like to report bad mount options here, but * traditionally we've ignored all mount options, so we'd * better continue to ignore non-existing options for bpf. */ - return opt == -ENOPARAM ? 0 : opt; + if (opt == -ENOPARAM) { + opt = vfs_parse_fs_param_source(fc, param); + if (opt != -ENOPARAM) + return opt; + + return 0; + } + + if (opt < 0) + return opt; + } switch (opt) { case OPT_MODE:

2 years, 11 months

1
0
0 0

FAILED: patch "[PATCH] net: dsa: mv88e6xxx: Add tx fwd offload PVT on intermediate" failed to apply to 5.16-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.16-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From e0068620e5e1779b3d5bb5649cd196e1cbf277a9 Mon Sep 17 00:00:00 2001 From: Tobias Waldekranz <tobias(a)waldekranz.com> Date: Thu, 9 Dec 2021 23:24:24 +0100 Subject: [PATCH] net: dsa: mv88e6xxx: Add tx fwd offload PVT on intermediate devices In a typical mv88e6xxx switch tree like this: CPU | .----. .--0--. | .--0--. | sw0 | | | sw1 | '-1-2-' | '-1-2-' '---' If sw1p{1,2} are added to a bridge that sw0p1 is not a part of, sw0 still needs to add a crosschip PVT entry for the virtual DSA device assigned to represent the bridge. Fixes: ce5df6894a57 ("net: dsa: mv88e6xxx: map virtual bridges with forwarding offload in the PVT") Signed-off-by: Tobias Waldekranz <tobias(a)waldekranz.com> Reviewed-by: Vladimir Oltean <vladimir.oltean(a)nxp.com> Signed-off-by: David S. Miller <davem(a)davemloft.net> diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c index 45aaa7333fb8..dced1505e6bf 100644 --- a/drivers/net/dsa/mv88e6xxx/chip.c +++ b/drivers/net/dsa/mv88e6xxx/chip.c @@ -2531,6 +2531,7 @@ static int mv88e6xxx_crosschip_bridge_join(struct dsa_switch *ds, mv88e6xxx_reg_lock(chip); err = mv88e6xxx_pvt_map(chip, sw_index, port); + err = err ? : mv88e6xxx_map_virtual_bridge_to_pvt(ds, bridge.num); mv88e6xxx_reg_unlock(chip); return err; @@ -2546,7 +2547,8 @@ static void mv88e6xxx_crosschip_bridge_leave(struct dsa_switch *ds, return; mv88e6xxx_reg_lock(chip); - if (mv88e6xxx_pvt_map(chip, sw_index, port)) + if (mv88e6xxx_pvt_map(chip, sw_index, port) || + mv88e6xxx_map_virtual_bridge_to_pvt(ds, bridge.num)) dev_err(ds->dev, "failed to remap cross-chip Port VLAN\n"); mv88e6xxx_reg_unlock(chip); }

2 years, 11 months

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror January 2022