| Message ID | 20210818002942.1607544-4-tony.luck@intel.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [v2,1/3] x86/mce: Avoid infinite loop for copy from user recovery | expand |
On Tue, Aug 17, 2021 at 05:29:42PM -0700, Tony Luck wrote:
> Fixes to the iterator code to handle faults
Can we name some of those fixes here pls?
We will forget which those were in the future so it would be good to
leave some breadcrumbs at least...
Thx.
> Can we name some of those fixes here pls?
Some/all of this bunch from Al Viro:
a180bd1d7e16 iov_iter: remove uaccess_kernel() warning from iov_iter_init()
6852df126699 csum_and_copy_to_pipe_iter(): leave handling of csum_state to caller
2a510a744beb clean up copy_mc_pipe_to_iter()
893839fd5733 pipe_zero(): we don't need no stinkin' kmap_atomic()...
2495bdcc86dc iov_iter: clean csum_and_copy_...() primitives up a bit
55ca375c5dcc copy_page_from_iter(): don't need kmap_atomic() for kvec/bvec cases
c1d4d6a9ae88 copy_page_to_iter(): don't bother with kmap_atomic() for bvec/kvec cases
4b179e9a9c7c iterate_xarray(): only of the first iteration we might get offset != 0
a6e4ec7bfd32 pull handling of ->iov_offset into iterate_{iovec,bvec,xarray}
7baa5099002f iov_iter: make iterator callbacks use base and len instead of iovec
622838f3fde2 iov_iter: make the amount already copied available to iterator callbacks
21b56c847753 iov_iter: get rid of separate bvec and xarray callbacks
1b4fb5ffd79b iov_iter: teach iterate_{bvec,xarray}() about possible short copies
7491a2bf64e3 iterate_bvec(): expand bvec.h macro forest, massage a bit
5c67aa90cd5c iov_iter: unify iterate_iovec and iterate_kvec
7a1bcb5d255d iov_iter: massage iterate_iovec and iterate_kvec to logics similar to iterate_bvec
f5da83545f4e iterate_and_advance(): get rid of magic in case when n is 0
594e450b3f44 csum_and_copy_to_iter(): massage into form closer to csum_and_copy_from_iter()
f0b65f39ac50 iov_iter: replace iov_iter_copy_from_user_atomic() with iterator-advancing variant
e4f8df86798a [xarray] iov_iter_npages(): just use DIV_ROUND_UP()
66531c65aa25 iov_iter_npages(): don't bother with iterate_all_kinds()
3d671ca62a08 get rid of iterate_all_kinds() in iov_iter_get_pages()/iov_iter_get_pages_alloc()
610c7a71543d iov_iter_gap_alignment(): get rid of iterate_all_kinds()
9221d2e37b72 iov_iter_alignment(): don't bother with iterate_all_kinds()
8409a0d261e2 sanitize iov_iter_fault_in_readable()
185ac4d43669 iov_iter: optimize iov_iter_advance() for iovec and kvec
8cd54c1c8480 iov_iter: separate direction from flavour
556351c1c09a iov_iter_advance(): don't modify ->iov_offset for ITER_DISCARD
28f38db7edbf iov_iter: reorder handling of flavours in primitives
4b6c132b7da6 iov_iter: switch ..._full() variants of primitives to use of iov_iter_revert()
3b3fc051cd2c iov_iter_advance(): use consistent semantics for move past the end
0e8f0d674015 [xarray] iov_iter_fault_in_readable() should do nothing in xarray case
a506abc7b644 copy_page_to_iter(): fix ITER_DISCARD case
08aa64796016 teach copy_page_to_iter() to handle compound pages
66cd071a1f83 iov_iter: Remove iov_iter_for_each_range()
-Tony
On Mon, Sep 20, 2021 at 04:18:58PM +0000, Luck, Tony wrote: > > Can we name some of those fixes here pls? > > Some/all of this bunch from Al Viro: Is this how you generated that list, per chance? $ git log --oneline v5.14 -- lib/iov_iter.c ? Output looks at least similar to what you've pasted...
> Is this how you generated that list, per chance? > > $ git log --oneline v5.14 -- lib/iov_iter.c Almost. I had "v5.14 ^v5.13" to stop git from going back further than when I think Al started applying those fixes. -Tony
diff --git a/arch/x86/lib/copy_user_64.S b/arch/x86/lib/copy_user_64.S index 57b79c577496..2797e630b9b1 100644 --- a/arch/x86/lib/copy_user_64.S +++ b/arch/x86/lib/copy_user_64.S @@ -234,24 +234,11 @@ EXPORT_SYMBOL(copy_user_enhanced_fast_string) */ SYM_CODE_START_LOCAL(.Lcopy_user_handle_tail) movl %edx,%ecx - cmp $X86_TRAP_MC,%eax /* check if X86_TRAP_MC */ - je 3f 1: rep movsb 2: mov %ecx,%eax ASM_CLAC ret - /* - * Return zero to pretend that this copy succeeded. This - * is counter-intuitive, but needed to prevent the code - * in lib/iov_iter.c from retrying and running back into - * the poison cache line again. The machine check handler - * will ensure that a SIGBUS is sent to the task. - */ -3: xorl %eax,%eax - ASM_CLAC - ret - _ASM_EXTABLE_CPY(1b, 2b) SYM_CODE_END(.Lcopy_user_handle_tail)
Fixes to the iterator code to handle faults that are not on page boundaries mean that the special case for machine check during copy from user is no longer needed. Signed-off-by: Tony Luck <tony.luck@intel.com> --- arch/x86/lib/copy_user_64.S | 13 ------------- 1 file changed, 13 deletions(-)