Cross Memory Attach v3
Subsystems (9)
PATCH RESEND 1/1 Cross Memory Attach v3
Hi Andrew,
This is essentially a resend of the same patch I sent a few months ago, just updated to v3.0-rc6. I'm hoping you have time to look at it now with the view of including it in your tree.
Just as a reminder of what has happened so far, repeating some content from previous emails about it:
The basic idea behind cross memory attach is to allow MPI programs doing intra-node communication to do a single copy of the message rather than a double copy of the message via shared memory.
The following patch attempts to achieve this by allowing a destination process, given an address and size from a source process, to copy memory directly from the source process into its own address space via a system call. There is also a symmetrical ability to copy from the current process's address space into a destination process's address space.
- Use of /proc/pid/mem has been considered, but there are issues with using it:
- Does not allow for specifying iovecs for both src and dest, assuming preadv or pwritev was implemented either the area read from or
written to would need to be contiguous.
- Currently mem_read allows only processes who are currently ptrace'ing the target and are still able to ptrace the target to read from the target. This check could possibly be moved to the open call, but its not clear exactly what race this restriction is stopping (reason appears to have been lost)
- Having to send the fd of /proc/self/mem via SCM_RIGHTS on unix domain socket is a bit ugly from a userspace point of view, especially when you may have hundreds if not (eventually) thousands of processes that all need to do this with each other
- Doesn't allow for some future use of the interface we would like to consider adding in the future (see below)
- Interestingly reading from /proc/pid/mem currently actually involves two copies! (But this could be fixed pretty easily)
As mentioned previously use of vmsplice instead was considered, but has problems. Since you need the reader and writer working co-operatively if the pipe is not drained then you block. Which requires some wrapping to do non blocking on the send side or polling on the receive. In all to all communication it requires ordering otherwise you can deadlock. And in the example of many MPI tasks writing to one MPI task vmsplice serialises the copying.
There are some cases of MPI collectives where even a single copy interface does not get us the performance gain we could. For example in an MPI_Reduce rather than copy the data from the source we would like to instead use it directly in a mathops (say the reduce is doing a sum) as this would save us doing a copy. We don't need to keep a copy of the data from the source. I haven't implemented this, but I think this interface could in the future do all this through the use of the flags
- eg could specify the math operation and type and the kernel rather than just copying the data would apply the specified operation between the source and destination and store it in the destination.
Although we don't have a "second user" of the interface (though I've had some nibbles from people who may be interested in using it for intra process messaging which is not MPI). This interface is something which hardware vendors are already doing for their custom drivers to implement fast local communication. And so in addition to this being useful for OpenMPI it would mean the driver maintainers don't have to fix things up when the mm changes.
There was some discussion about how much faster a true zero copy would go. Here's a link back to the email with some testing I did on that:
http://marc.info/?l=linux-mm&m=130105930902915&w=2
Regards,
Chris
Signature · 18 lines
[email protected] Signed-off-by: Chris Yeoh <[email protected]> --- arch/powerpc/include/asm/systbl.h | 2 arch/powerpc/include/asm/unistd.h | 4 arch/x86/include/asm/unistd_32.h | 4 arch/x86/kernel/syscall_table_32.S | 2 fs/aio.c | 4 fs/compat.c | 7 fs/read_write.c | 8 include/linux/compat.h | 3 include/linux/fs.h | 7 include/linux/syscalls.h | 13 + mm/Makefile | 3 mm/process_vm_access.c | 446 +++++++++++++++++++++++++++++++++++++ security/keys/compat.c | 2 security/keys/keyctl.c | 2 14 files changed, 490 insertions(+), 17 deletions(-)
arch/powerpc/include/asm/systbl.h +2 -0
@@ -354,3 +354,5 @@ COMPAT_SYS_SPU(clock_adjtime) SYSCALL_SPU(syncfs) COMPAT_SYS_SPU(sendmmsg) SYSCALL_SPU(setns) +COMPAT_SYS(process_vm_readv) +COMPAT_SYS(process_vm_writev)
arch/powerpc/include/asm/unistd.h +3 -1
@@ -373,10 +373,12 @@#define __NR_syncfs 348#define __NR_sendmmsg 349#define __NR_setns 350 +#define __NR_process_vm_readv 351 +#define __NR_process_vm_writev 352#ifdef __KERNEL__ -#define __NR_syscalls 351 +#define __NR_syscalls 353#define __NR__exit __NR_exit#define NR_syscalls __NR_syscalls
arch/x86/include/asm/unistd_32.h +3 -1
@@ -352,10 +352,12 @@#define __NR_syncfs 344#define __NR_sendmmsg 345#define __NR_setns 346 +#define __NR_process_vm_readv 347 +#define __NR_process_vm_writev 348#ifdef __KERNEL__ -#define NR_syscalls 347 +#define NR_syscalls 349#define __ARCH_WANT_IPC_PARSE_VERSION#define __ARCH_WANT_OLD_READDIR
arch/x86/kernel/syscall_table_32.S +2 -0
@@ -346,3 +346,5 @@ ENTRY(sys_call_table) .long sys_syncfs .long sys_sendmmsg /* 345 */ .long sys_setns + .long sys_process_vm_readv + .long sys_process_vm_writev
fs/aio.c +2 -2
@@ -1387,13 +1387,13 @@ static ssize_t aio_setup_vectored_rw(int type, struct kiocb *kiocb, bool compat) ret =compat_rw_copy_check_uvector (type, (struct compat_iovec __user *)kiocb->ki_buf, kiocb->ki_nbytes,1 , &kiocb->ki_inline_vec, - &kiocb->ki_iovec); + &kiocb->ki_iovec,1 ); else#endif ret =rw_copy_check_uvector (type, (struct iovec __user *)kiocb->ki_buf, kiocb->ki_nbytes,1 , &kiocb->ki_inline_vec, - &kiocb->ki_iovec); + &kiocb->ki_iovec,1 );if (ret <0 ) goto out;
fs/compat.c +4 -3
@@ -550,7 +550,7 @@ out:ssize_t compat_rw_copy_check_uvector (int type,const struct compat_iovec __user *uvector,unsigned long nr_segs,unsigned long fast_segs,struct iovec *fast_pointer, -struct iovec **ret_pointer) +struct iovec **ret_pointer,int check_access) {compat_ssize_t tot_len;struct iovec *iov = *ret_pointer = fast_pointer;
@@ -597,7 +597,8 @@ ssize_t compat_rw_copy_check_uvector(int type, }if (len <0 )/* size_t not fitting in compat_ssize_t .. */ goto out; -if (!access_ok (vrfy_dir (type),compat_ptr (buf), len)) { +if (check_access && + !access_ok (vrfy_dir (type),compat_ptr (buf), len)) { ret = -EFAULT ; goto out; }
@@ -1111,7 +1112,7 @@ static ssize_t compat_do_readv_writev(int type, struct file *file, goto out; tot_len =compat_rw_copy_check_uvector (type, uvector, nr_segs, -UIO_FASTIOV , iovstack, &iov); +UIO_FASTIOV , iovstack, &iov,1 );if (tot_len ==0 ) { ret =0 ; goto out;
fs/read_write.c +5 -3
@@ -575,7 +575,8 @@ ssize_t do_loop_readv_writev(struct file *filp, struct iovec *iov,ssize_t rw_copy_check_uvector (int type,const struct iovec __user * uvector,unsigned long nr_segs,unsigned long fast_segs,struct iovec *fast_pointer, -struct iovec **ret_pointer) +struct iovec **ret_pointer, +int check_access) {unsigned long seg;ssize_t ret;
@@ -631,7 +632,8 @@ ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, ret = -EINVAL ; goto out; } -if (unlikely (!access_ok (vrfy_dir (type), buf, len))) { +if (check_access + &&unlikely (!access_ok (vrfy_dir (type), buf, len))) { ret = -EFAULT ; goto out; }
@@ -663,7 +665,7 @@ static ssize_t do_readv_writev(int type, struct file *file, } ret =rw_copy_check_uvector (type, uvector, nr_segs, -ARRAY_SIZE (iovstack), iovstack, &iov); +ARRAY_SIZE (iovstack), iovstack, &iov,1 );if (ret <=0 ) goto out;
include/linux/compat.h +2 -1
@@ -557,7 +557,8 @@ extern ssize_t compat_rw_copy_check_uvector(int type,const struct compat_iovec __user *uvector, unsigned long nr_segs, unsigned long fast_segs, struct iovec *fast_pointer, - struct iovec **ret_pointer); + struct iovec **ret_pointer, + int check_access);extern void __user *compat_alloc_user_space (unsigned long len);
include/linux/fs.h +4 -3
@@ -1608,9 +1608,10 @@ struct inode_operations {struct seq_file ;ssize_t rw_copy_check_uvector (int type,const struct iovec __user * uvector, -unsigned long nr_segs,unsigned long fast_segs, -struct iovec *fast_pointer, -struct iovec **ret_pointer); +unsigned long nr_segs,unsigned long fast_segs, +struct iovec *fast_pointer, +struct iovec **ret_pointer, +int check_access);extern ssize_t vfs_read (struct file *,char __user *,size_t ,loff_t *);extern ssize_t vfs_write (struct file *,const char __user *,size_t ,loff_t *);
include/linux/syscalls.h +13 -0
@@ -847,4 +847,17 @@ asmlinkage long sys_open_by_handle_at(int mountdirfd,struct file_handle __user *handle, int flags);asmlinkage long sys_setns (int fd,int nstype); +asmlinkage long sys_process_vm_readv (pid_t pid, +const struct iovec __user *lvec, +unsigned long liovcnt, +const struct iovec __user *rvec, +unsigned long riovcnt, +unsigned long flags); +asmlinkage long sys_process_vm_writev (pid_t pid, +const struct iovec __user *lvec, +unsigned long liovcnt, +const struct iovec __user *rvec, +unsigned long riovcnt, +unsigned long flags); +#endif
mm/Makefile +2 -1
@@ -5,7 +5,8 @@ mmu-y := nommu.o mmu-$(CONFIG_MMU) := fremap.o highmem.o madvise.o memory.o mincore.o \ mlock.o mmap.o mprotect.o mremap.o msync.o rmap.o \ - vmalloc.o pagewalk.o pgtable-generic.o + vmalloc.o pagewalk.o pgtable-generic.o \ + process_vm_access.o obj-y := filemap.o mempool.o oom_kill.o fadvise.o \ maccess.o page_alloc.o page-writeback.o \
mm/process_vm_access.c new file +446 -0
446 lines not on this page · raw message
security/keys/compat.c +1 -1
@@ -38,7 +38,7 @@ long compat_keyctl_instantiate_key_iov( ret =compat_rw_copy_check_uvector (WRITE , _payload_iov, ioc,ARRAY_SIZE (iovstack), - iovstack, &iov); + iovstack, &iov,1 );if (ret <0 )return ret;if (ret ==0 )
security/keys/keyctl.c +1 -1
@@ -1065,7 +1065,7 @@ long keyctl_instantiate_key_iov(key_serial_t id, goto no_payload; ret =rw_copy_check_uvector (WRITE , _payload_iov, ioc, -ARRAY_SIZE (iovstack), iovstack, &iov); +ARRAY_SIZE (iovstack), iovstack, &iov,1 );if (ret <0 )return ret;if (ret ==0 )
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Andrew Morton
maintainermm/process_vm_access.c line 273
Quoted 10 lines · On Fri, 8 Jul 2011 18:06:07 +0930 Christopher Yeoh <[email protected]> wrote:
mm/process_vm_access.c new file +7
-0+static ssize_t process_vm_rw (pid_t pid,const struct iovec *lvec, +unsigned long liovcnt, +const struct iovec *rvec, +unsigned long riovcnt, +unsigned long flags,int vm_write) +{ +if (!mm || (task->flags &PF_KTHREAD )) {
Can a PF_KTHREAD thread have a non-zero ->mm?
mm/process_vm_access.c line 277
Quoted 4 lines
mm/process_vm_access.c new file +4
-0+task_unlock (task); + rc = -EINVAL ; + goto put_task_struct; + }
anyway, grumble.
Please resend, cc'ing linux-kernel.
Please also cc [email protected] and describe (within the changelog context) the plan for getting the manpages written.
Please also cc [email protected] and provide (within the changelog context) arch developers with the documentation and test code which will enable them to implement this and to verify it with minimal effort.
Thanks.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Quoted 12 lines · On Fri, 15 Jul 2011 15:37:43 -0700 Andrew Morton <[email protected]>…
Quoted 9 lines · On Fri, 8 Jul 2011 18:06:07 +0930 Christopher Yeoh <[email protected]> wrote:
mm/process_vm_access.c new file +6
-0+static ssize_t process_vm_rw (pid_t pid,const struct iovec *lvec, +unsigned long liovcnt, +const struct iovec *rvec, +unsigned long riovcnt, +unsigned long flags,int vm_write) +{
Quoted 7 lines
mm/process_vm_access.c line 273
Quoted 1 line
mm/process_vm_access.c new file +1
-0+if (!mm || (task->flags &PF_KTHREAD )) {Can a
PF_KTHREADthread have a non-zero ->mm?mm/process_vm_access.c line 277
Quoted 4 lines
mm/process_vm_access.c new file +4
-0+task_unlock (task); + rc = -EINVAL ; + goto put_task_struct; + }
According to get_task_mm it can:
/**
* get_task_mm - acquire a reference to the task's mm
*
* Returns %NULL if the task has no mm. Checks PF_KTHREAD (meaning
* this kernel workthread has transiently adopted a user mm with use_mm,
* to do its AIO) is not set and if so returns a reference to it, after
* bumping up the use count. User must release the mm via mmput()
* after use. Typically used by /proc and ptrace.
*/
Quoted 3 lines
anyway, grumble.
Please resend, cc'ing linux-kernel.
Am doing the CC resends in a separate email...
Chris
Signature · 8 lines
[email protected] -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"[email protected]"> [email protected] </a>