From: Damien Le Moal damien.lemoal@opensource.wdc.com

commit c1c1204c0d0c1dccc1310b9277fb2bd8b663d8fe upstream.

If a readahead is issued to a sequential zone file with an offset exactly equal to the current file size, the iomap type is set to IOMAP_UNWRITTEN, which will prevent an IO, but the iomap length is calculated as 0. This causes a WARN_ON() in iomap_iter():

[17309.548939] WARNING: CPU: 3 PID: 2137 at fs/iomap/iter.c:34 iomap_iter+0x9cf/0xe80 [...] [17309.650907] RIP: 0010:iomap_iter+0x9cf/0xe80 [...] [17309.754560] Call Trace: [17309.757078] <TASK> [17309.759240] ? lock_is_held_type+0xd8/0x130 [17309.763531] iomap_readahead+0x1a8/0x870 [17309.767550] ? iomap_read_folio+0x4c0/0x4c0 [17309.771817] ? lockdep_hardirqs_on_prepare+0x400/0x400 [17309.778848] ? lock_release+0x370/0x750 [17309.784462] ? folio_add_lru+0x217/0x3f0 [17309.790220] ? reacquire_held_locks+0x4e0/0x4e0 [17309.796543] read_pages+0x17d/0xb60 [17309.801854] ? folio_add_lru+0x238/0x3f0 [17309.807573] ? readahead_expand+0x5f0/0x5f0 [17309.813554] ? policy_node+0xb5/0x140 [17309.819018] page_cache_ra_unbounded+0x27d/0x450 [17309.825439] filemap_get_pages+0x500/0x1450 [17309.831444] ? filemap_add_folio+0x140/0x140 [17309.837519] ? lock_is_held_type+0xd8/0x130 [17309.843509] filemap_read+0x28c/0x9f0 [17309.848953] ? zonefs_file_read_iter+0x1ea/0x4d0 [zonefs] [17309.856162] ? trace_contention_end+0xd6/0x130 [17309.862416] ? __mutex_lock+0x221/0x1480 [17309.868151] ? zonefs_file_read_iter+0x166/0x4d0 [zonefs] [17309.875364] ? filemap_get_pages+0x1450/0x1450 [17309.881647] ? __mutex_unlock_slowpath+0x15e/0x620 [17309.888248] ? wait_for_completion_io_timeout+0x20/0x20 [17309.895231] ? lock_is_held_type+0xd8/0x130 [17309.901115] ? lock_is_held_type+0xd8/0x130 [17309.906934] zonefs_file_read_iter+0x356/0x4d0 [zonefs] [17309.913750] new_sync_read+0x2d8/0x520 [17309.919035] ? __x64_sys_lseek+0x1d0/0x1d0

Furthermore, this causes iomap_readahead() to loop forever as iomap_readahead_iter() always returns 0, making no progress.

Fix this by treating reads after the file size as access to holes, setting the iomap type to IOMAP_HOLE, the iomap addr to IOMAP_NULL_ADDR and using the length argument as is for the iomap length. To simplify the code with this change, zonefs_iomap_begin() is split into the read variant, zonefs_read_iomap_begin() and zonefs_read_iomap_ops, and the write variant, zonefs_write_iomap_begin() and zonefs_write_iomap_ops.

Reported-by: Jorgen Hansen Jorgen.Hansen@wdc.com Fixes: 8dcc1a9d90c1 ("fs: New zonefs file system") Signed-off-by: Damien Le Moal damien.lemoal@opensource.wdc.com Reviewed-by: Christoph Hellwig hch@lst.de Reviewed-by: Johannes Thumshirn johannes.thumshirn@wdc.com Reviewed-by: Jorgen Hansen Jorgen.Hansen@wdc.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- fs/zonefs/super.c | 92 ++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 62 insertions(+), 30 deletions(-)

--- a/fs/zonefs/super.c +++ b/fs/zonefs/super.c @@ -68,15 +68,49 @@ static inline void zonefs_i_size_write(s zi->i_flags &= ~ZONEFS_ZONE_OPEN; }

-static int zonefs_iomap_begin(struct inode *inode, loff_t offset, loff_t length, - unsigned int flags, struct iomap *iomap, - struct iomap *srcmap) +static int zonefs_read_iomap_begin(struct inode *inode, loff_t offset, + loff_t length, unsigned int flags, + struct iomap *iomap, struct iomap *srcmap) { struct zonefs_inode_info *zi = ZONEFS_I(inode); struct super_block *sb = inode->i_sb; loff_t isize;

- /* All I/Os should always be within the file maximum size */ + /* + * All blocks are always mapped below EOF. If reading past EOF, + * act as if there is a hole up to the file maximum size. + */ + mutex_lock(&zi->i_truncate_mutex); + iomap->bdev = inode->i_sb->s_bdev; + iomap->offset = ALIGN_DOWN(offset, sb->s_blocksize); + isize = i_size_read(inode); + if (iomap->offset >= isize) { + iomap->type = IOMAP_HOLE; + iomap->addr = IOMAP_NULL_ADDR; + iomap->length = length; + } else { + iomap->type = IOMAP_MAPPED; + iomap->addr = (zi->i_zsector << SECTOR_SHIFT) + iomap->offset; + iomap->length = isize - iomap->offset; + } + mutex_unlock(&zi->i_truncate_mutex); + + return 0; +} + +static const struct iomap_ops zonefs_read_iomap_ops = { + .iomap_begin = zonefs_read_iomap_begin, +}; + +static int zonefs_write_iomap_begin(struct inode *inode, loff_t offset, + loff_t length, unsigned int flags, + struct iomap *iomap, struct iomap *srcmap) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct super_block *sb = inode->i_sb; + loff_t isize; + + /* All write I/Os should always be within the file maximum size */ if (WARN_ON_ONCE(offset + length > zi->i_max_size)) return -EIO;

@@ -86,7 +120,7 @@ static int zonefs_iomap_begin(struct ino * operation. */ if (WARN_ON_ONCE(zi->i_ztype == ZONEFS_ZTYPE_SEQ && - (flags & IOMAP_WRITE) && !(flags & IOMAP_DIRECT))) + !(flags & IOMAP_DIRECT))) return -EIO;

/* @@ -95,45 +129,42 @@ static int zonefs_iomap_begin(struct ino * write pointer) and unwriten beyond. */ mutex_lock(&zi->i_truncate_mutex); + iomap->bdev = inode->i_sb->s_bdev; + iomap->offset = ALIGN_DOWN(offset, sb->s_blocksize); + iomap->addr = (zi->i_zsector << SECTOR_SHIFT) + iomap->offset; isize = i_size_read(inode); - if (offset >= isize) + if (iomap->offset >= isize) { iomap->type = IOMAP_UNWRITTEN; - else + iomap->length = zi->i_max_size - iomap->offset; + } else { iomap->type = IOMAP_MAPPED; - if (flags & IOMAP_WRITE) - length = zi->i_max_size - offset; - else - length = min(length, isize - offset); + iomap->length = isize - iomap->offset; + } mutex_unlock(&zi->i_truncate_mutex);

- iomap->offset = ALIGN_DOWN(offset, sb->s_blocksize); - iomap->length = ALIGN(offset + length, sb->s_blocksize) - iomap->offset; - iomap->bdev = inode->i_sb->s_bdev; - iomap->addr = (zi->i_zsector << SECTOR_SHIFT) + iomap->offset; - return 0; }

-static const struct iomap_ops zonefs_iomap_ops = { - .iomap_begin = zonefs_iomap_begin, +static const struct iomap_ops zonefs_write_iomap_ops = { + .iomap_begin = zonefs_write_iomap_begin, };

static int zonefs_readpage(struct file *unused, struct page *page) { - return iomap_readpage(page, &zonefs_iomap_ops); + return iomap_readpage(page, &zonefs_read_iomap_ops); }

static void zonefs_readahead(struct readahead_control *rac) { - iomap_readahead(rac, &zonefs_iomap_ops); + iomap_readahead(rac, &zonefs_read_iomap_ops); }

/* * Map blocks for page writeback. This is used only on conventional zone files, * which implies that the page range can only be within the fixed inode size. */ -static int zonefs_map_blocks(struct iomap_writepage_ctx *wpc, - struct inode *inode, loff_t offset) +static int zonefs_write_map_blocks(struct iomap_writepage_ctx *wpc, + struct inode *inode, loff_t offset) { struct zonefs_inode_info *zi = ZONEFS_I(inode);

@@ -147,12 +178,12 @@ static int zonefs_map_blocks(struct ioma offset < wpc->iomap.offset + wpc->iomap.length) return 0;

- return zonefs_iomap_begin(inode, offset, zi->i_max_size - offset, - IOMAP_WRITE, &wpc->iomap, NULL); + return zonefs_write_iomap_begin(inode, offset, zi->i_max_size - offset, + IOMAP_WRITE, &wpc->iomap, NULL); }

static const struct iomap_writeback_ops zonefs_writeback_ops = { - .map_blocks = zonefs_map_blocks, + .map_blocks = zonefs_write_map_blocks, };

static int zonefs_writepage(struct page *page, struct writeback_control *wbc) @@ -182,7 +213,8 @@ static int zonefs_swap_activate(struct s return -EINVAL; }

- return iomap_swapfile_activate(sis, swap_file, span, &zonefs_iomap_ops); + return iomap_swapfile_activate(sis, swap_file, span, + &zonefs_read_iomap_ops); }

static const struct address_space_operations zonefs_file_aops = { @@ -612,7 +644,7 @@ static vm_fault_t zonefs_filemap_page_mk

/* Serialize against truncates */ down_read(&zi->i_mmap_sem); - ret = iomap_page_mkwrite(vmf, &zonefs_iomap_ops); + ret = iomap_page_mkwrite(vmf, &zonefs_write_iomap_ops); up_read(&zi->i_mmap_sem);

sb_end_pagefault(inode->i_sb); @@ -869,7 +901,7 @@ static ssize_t zonefs_file_dio_write(str if (append) ret = zonefs_file_dio_append(iocb, from); else - ret = iomap_dio_rw(iocb, from, &zonefs_iomap_ops, + ret = iomap_dio_rw(iocb, from, &zonefs_write_iomap_ops, &zonefs_write_dio_ops, sync); if (zi->i_ztype == ZONEFS_ZTYPE_SEQ && (ret > 0 || ret == -EIOCBQUEUED)) { @@ -911,7 +943,7 @@ static ssize_t zonefs_file_buffered_writ if (ret <= 0) goto inode_unlock;

- ret = iomap_file_buffered_write(iocb, from, &zonefs_iomap_ops); + ret = iomap_file_buffered_write(iocb, from, &zonefs_write_iomap_ops); if (ret > 0) iocb->ki_pos += ret; else if (ret == -EIO) @@ -1004,7 +1036,7 @@ static ssize_t zonefs_file_read_iter(str goto inode_unlock; } file_accessed(iocb->ki_filp); - ret = iomap_dio_rw(iocb, to, &zonefs_iomap_ops, + ret = iomap_dio_rw(iocb, to, &zonefs_read_iomap_ops, &zonefs_read_dio_ops, is_sync_kiocb(iocb)); } else { ret = generic_file_read_iter(iocb, to);

From: Eric Dumazet edumazet@google.com

commit c579bd1b4021c42ae247108f1e6f73dd3f08600c upstream.

Even when implementing RFC 6056 3.3.4 (Algorithm 4: Double-Hash Port Selection Algorithm), a patient attacker could still be able to collect enough state from an otherwise idle host.

Idea of this patch is to inject some noise, in the cases __inet_hash_connect() found a candidate in the first attempt.

This noise should not significantly reduce the collision avoidance, and should be zero if connection table is already well used.

Note that this is not implementing RFC 6056 3.3.5 because we think Algorithm 5 could hurt typical workloads.

Signed-off-by: Eric Dumazet edumazet@google.com Cc: David Dworken ddworken@google.com Cc: Willem de Bruijn willemb@google.com Signed-off-by: David S. Miller davem@davemloft.net Cc: Ben Hutchings ben@decadent.org.uk Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- net/ipv4/inet_hashtables.c | 5 +++++ 1 file changed, 5 insertions(+)

--- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -833,6 +833,11 @@ next_port: return -EADDRNOTAVAIL;

ok: + /* If our first attempt found a candidate, skip next candidate + * in 1/16 of cases to add some noise. + */ + if (!i && !(prandom_u32() % 16)) + i = 2; WRITE_ONCE(table_perturb[index], READ_ONCE(table_perturb[index]) + i + 2);

/* Head lock still held and bh's disabled */

From: Willy Tarreau w@1wt.eu

commit ca7af0402550f9a0b3316d5f1c30904e42ed257d upstream.

Here we're randomly adding between 0 and 7 random increments to the selected source port in order to add some noise in the source port selection that will make the next port less predictable.

With the default port range of 32768-60999 this means a worst case reuse scenario of 14116/8=1764 connections between two consecutive uses of the same port, with an average of 14116/4.5=3137. This code was stressed at more than 800000 connections per second to a fixed target with all connections closed by the client using RSTs (worst condition) and only 2 connections failed among 13 billion, despite the hash being reseeded every 10 seconds, indicating a perfectly safe situation.

Cc: Moshe Kol moshe.kol@mail.huji.ac.il Cc: Yossi Gilad yossi.gilad@mail.huji.ac.il Cc: Amit Klein aksecurity@gmail.com Reviewed-by: Eric Dumazet edumazet@google.com Signed-off-by: Willy Tarreau w@1wt.eu Signed-off-by: Jakub Kicinski kuba@kernel.org Cc: Ben Hutchings ben@decadent.org.uk Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- net/ipv4/inet_hashtables.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-)

--- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -833,11 +833,12 @@ next_port: return -EADDRNOTAVAIL;

ok: - /* If our first attempt found a candidate, skip next candidate - * in 1/16 of cases to add some noise. + /* Here we want to add a little bit of randomness to the next source + * port that will be chosen. We use a max() with a random here so that + * on low contention the randomness is maximal and on high contention + * it may be inexistent. */ - if (!i && !(prandom_u32() % 16)) - i = 2; + i = max_t(int, i, (prandom_u32() & 7) * 2); WRITE_ONCE(table_perturb[index], READ_ONCE(table_perturb[index]) + i + 2);

/* Head lock still held and bh's disabled */

From: Willy Tarreau w@1wt.eu

commit 4c2c8f03a5ab7cb04ec64724d7d176d00bcc91e5 upstream.

Moshe Kol, Amit Klein, and Yossi Gilad reported being able to accurately identify a client by forcing it to emit only 40 times more connections than there are entries in the table_perturb[] table. The previous two improvements consisting in resalting the secret every 10s and adding randomness to each port selection only slightly improved the situation, and the current value of 2^8 was too small as it's not very difficult to make a client emit 10k connections in less than 10 seconds.

Thus we're increasing the perturb table from 2^8 to 2^16 so that the same precision now requires 2.6M connections, which is more difficult in this time frame and harder to hide as a background activity. The impact is that the table now uses 256 kB instead of 1 kB, which could mostly affect devices making frequent outgoing connections. However such components usually target a small set of destinations (load balancers, database clients, perf assessment tools), and in practice only a few entries will be visited, like before.

A live test at 1 million connections per second showed no performance difference from the previous value.

Reported-by: Moshe Kol moshe.kol@mail.huji.ac.il Reported-by: Yossi Gilad yossi.gilad@mail.huji.ac.il Reported-by: Amit Klein aksecurity@gmail.com Reviewed-by: Eric Dumazet edumazet@google.com Signed-off-by: Willy Tarreau w@1wt.eu Signed-off-by: Jakub Kicinski kuba@kernel.org Cc: Ben Hutchings ben@decadent.org.uk Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- net/ipv4/inet_hashtables.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-)

--- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -726,11 +726,12 @@ EXPORT_SYMBOL_GPL(inet_unhash); * Note that we use 32bit integers (vs RFC 'short integers') * because 2^16 is not a multiple of num_ephemeral and this * property might be used by clever attacker. - * RFC claims using TABLE_LENGTH=10 buckets gives an improvement, - * we use 256 instead to really give more isolation and - * privacy, this only consumes 1 KB of kernel memory. + * RFC claims using TABLE_LENGTH=10 buckets gives an improvement, though + * attacks were since demonstrated, thus we use 65536 instead to really + * give more isolation and privacy, at the expense of 256kB of kernel + * memory. */ -#define INET_TABLE_PERTURB_SHIFT 8 +#define INET_TABLE_PERTURB_SHIFT 16 #define INET_TABLE_PERTURB_SIZE (1 << INET_TABLE_PERTURB_SHIFT) static u32 *table_perturb;

From: Lukas Wunner lukas@wunner.de

commit 2dd8a74fddd21b95dcc60a2d3c9eaec993419d69 upstream.

RTS polarity of rs485-enabled ports is currently initialized on uart open via:

tty_port_open() tty_port_block_til_ready() tty_port_raise_dtr_rts() # if (C_BAUD(tty)) uart_dtr_rts() uart_port_dtr_rts()

There's at least three problems here:

First, if no baud rate is set, RTS polarity is not initialized. That's the right thing to do for rs232, but not for rs485, which requires that RTS is deasserted unconditionally.

Second, if the DeviceTree property "linux,rs485-enabled-at-boot-time" is present, RTS should be deasserted as early as possible, i.e. on probe. Otherwise it may remain asserted until first open.

Third, even though RTS is deasserted on open and close, it may subsequently be asserted by uart_throttle(), uart_unthrottle() or uart_set_termios() because those functions aren't rs485-aware. (Only uart_tiocmset() is.)

To address these issues, move RTS initialization from uart_port_dtr_rts() to uart_configure_port(). Prevent subsequent modification of RTS polarity by moving the existing rs485 check from uart_tiocmget() to uart_update_mctrl().

That way, RTS is initialized on probe and then remains unmodified unless the uart transmits data. If rs485 is enabled at runtime (instead of at boot) through a TIOCSRS485 ioctl(), RTS is initialized by the uart driver's ->rs485_config() callback and then likewise remains unmodified.

The PL011 driver initializes RTS on uart open and prevents subsequent modification in its ->set_mctrl() callback. That code is obsoleted by the present commit, so drop it.

Cc: Jan Kiszka jan.kiszka@siemens.com Cc: Su Bao Cheng baocheng.su@siemens.com Signed-off-by: Lukas Wunner lukas@wunner.de Link: https://lore.kernel.org/r/2d2acaf3a69e89b7bf687c912022b11fd29dfa1e.164290928... Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- drivers/tty/serial/serial_core.c | 34 ++++++++++++---------------------- 1 file changed, 12 insertions(+), 22 deletions(-)

--- a/drivers/tty/serial/serial_core.c +++ b/drivers/tty/serial/serial_core.c @@ -144,6 +144,11 @@ uart_update_mctrl(struct uart_port *port unsigned long flags; unsigned int old;

+ if (port->rs485.flags & SER_RS485_ENABLED) { + set &= ~TIOCM_RTS; + clear &= ~TIOCM_RTS; + } + spin_lock_irqsave(&port->lock, flags); old = port->mctrl; port->mctrl = (old & ~clear) | set; @@ -157,23 +162,10 @@ uart_update_mctrl(struct uart_port *port

static void uart_port_dtr_rts(struct uart_port *uport, int raise) { - int rs485_on = uport->rs485_config && - (uport->rs485.flags & SER_RS485_ENABLED); - int RTS_after_send = !!(uport->rs485.flags & SER_RS485_RTS_AFTER_SEND); - - if (raise) { - if (rs485_on && RTS_after_send) { - uart_set_mctrl(uport, TIOCM_DTR); - uart_clear_mctrl(uport, TIOCM_RTS); - } else { - uart_set_mctrl(uport, TIOCM_DTR | TIOCM_RTS); - } - } else { - unsigned int clear = TIOCM_DTR; - - clear |= (!rs485_on || RTS_after_send) ? TIOCM_RTS : 0; - uart_clear_mctrl(uport, clear); - } + if (raise) + uart_set_mctrl(uport, TIOCM_DTR | TIOCM_RTS); + else + uart_clear_mctrl(uport, TIOCM_DTR | TIOCM_RTS); }

/* @@ -1116,11 +1108,6 @@ uart_tiocmset(struct tty_struct *tty, un goto out;

if (!tty_io_error(tty)) { - if (uport->rs485.flags & SER_RS485_ENABLED) { - set &= ~TIOCM_RTS; - clear &= ~TIOCM_RTS; - } - uart_update_mctrl(uport, set, clear); ret = 0; } @@ -2429,6 +2416,9 @@ uart_configure_port(struct uart_driver * */ spin_lock_irqsave(&port->lock, flags); port->mctrl &= TIOCM_DTR; + if (port->rs485.flags & SER_RS485_ENABLED && + !(port->rs485.flags & SER_RS485_RTS_AFTER_SEND)) + port->mctrl |= TIOCM_RTS; port->ops->set_mctrl(port, port->mctrl); spin_unlock_irqrestore(&port->lock, flags);