uprobes && shmem (Was: uprobes: Shift ->readpage check from __copy_insn() to uprobe_register())
Hi Hugh,
Quoted 2 lines · Oleg Nesterov
copy_insn()fails with-EIOif ->readpage == NULL
In particular, this means that we can not probe the binaries on tmpfs. This is pity.
It seems that the potential fix is trivial, copy_insn() could use shmem_getpage_gfp(). But, is there any way to figure out that this inode/mapping/aops/whatever is actually shmem?
I am looking at shmem_get_inode() and I see nothing which could help, and shmem_aops/etc are all static.
Oleg.
Quoted 6 lines · Oleg Nesterov
Quoted 2 lines · Oleg Nesterov
copy_insn()fails with-EIOif ->readpage == NULLIn particular, this means that we can not probe the binaries on tmpfs. This is pity.
Yes, that is a pity: thanks for noticing.
Quoted 3 lines
It seems that the potential fix is trivial,
copy_insn()could useshmem_getpage_gfp(). But, is there any way to figure out that this
shmem_getpage_gfp() itself is static: please use shmem_read_mapping_page(mapping, pgoff): inline in linux/shmem_fs.h, calls shmem_read_mapping_page_gfp() in mm/shmem.c (a very few places need to override gfp_mask too: you do not), calls shmem_getpage_gfp().
Quoted 4 lines
inode/mapping/aops/whatever is actually shmem?
I am looking at
shmem_get_inode()and I see nothing which could help, and shmem_aops/etc are all static.
On 3.15 and later, you're in luck: Hannes added bool shmem_mapping(mapping) in his 0cd6144aadd2 "mm + fs: prepare for non-page entries in page cache radix trees"; and I just checked, it builds for "tiny" !CONFIG_SHMEM too.
If you're backporting to an earlier kernel, it would probably be best to add in a very small patch, extracting just shmem_mapping() and its linux/mm.h declarations from 0cd6144aadd2.
I notice shmem_mapping() checks backing_dev_info,
whereas shmem_get_mapping_page_gfp() checks a_ops: no problem in that. But it reminds me that you should test uprobe on SysV SHM when you're
done: again I think no problem, but there's an incestuous relationship between shm and shmem that can catch us out when adding such checks.
Hugh
Quoted 10 lines · Hugh Dickins
Quoted 6 lines · Oleg Nesterov
Quoted 2 lines · Oleg Nesterov
copy_insn()fails with-EIOif ->readpage == NULLIn particular, this means that we can not probe the binaries on tmpfs. This is pity.
Yes, that is a pity: thanks for noticing.
Thanks to Denys ;)
Quoted 7 lines
Quoted 2 lines
It seems that the potential fix is trivial,
copy_insn()could useshmem_getpage_gfp(). But, is there any way to figure out that this
shmem_getpage_gfp()itself is static: please useshmem_read_mapping_page(mapping, pgoff): inline inlinux/shmem_fs.h, callsshmem_read_mapping_page_gfp()inmm/shmem.c(a very few places need to overridegfp_masktoo: you do not), callsshmem_getpage_gfp().
Even better, thanks,
Quoted 8 lines
Quoted 4 lines
inode/mapping/aops/whatever is actually shmem?
I am looking at
shmem_get_inode()and I see nothing which could help, and shmem_aops/etc are all static.On 3.15 and later, you're in luck: Hannes added bool
shmem_mapping(mapping)in his0cd6144aadd2"mm + fs: prepare for non-page entries in page cache radix trees"; and I just checked, it builds for "tiny" !CONFIG_SHMEM too.
Heh. I need to do git pull more often, I guess. Great.
Quoted 9 lines
If you're backporting to an earlier kernel, it would probably be best to add in a very small patch, extracting just
shmem_mapping()and itslinux/mm.hdeclarations from0cd6144aadd2.I notice
shmem_mapping()checksbacking_dev_info,
whereasshmem_get_mapping_page_gfp()checksa_ops: no problem in that. But it reminds me that you should test uprobe on SysV SHM when you're
done: again I think no problem, but there's an incestuous relationship between shm and shmem that can catch us out when adding such checks.
Hugh, thanks a lot!
Oleg.
Quoted 11 lines · Hugh Dickins
Quoted 2 lines · Oleg Nesterov
shmem_getpage_gfp(). But, is there any way to figure out that this inode/mapping/aops/whatever is actually shmem?On 3.15 and later, you're in luck: Hannes added bool
shmem_mapping(mapping)in his0cd6144aadd2"mm + fs: prepare for non-page entries in page cache radix trees"; and I just checked, it builds for "tiny" !CONFIG_SHMEM too.If you're backporting to an earlier kernel, it would probably be best to add in a very small patch, extracting just
shmem_mapping()and itslinux/mm.hdeclarations from0cd6144aadd2.
Looking into something else, I've just been reminded of mapping_cap_swap_backed(mapping): mm/madvise.c has been using that test for a year. It also returns true on swapper_space, the imaginary mapping you get from a PageSwapCache page: so I suppose it would be wrong to make Hannes's shmem_mapping() a wrapper to that. But for any backport of your change, it would better than extracting part of 0cd6144aadd2.
(Sorry for going on about backporting, when you may have no such intention.)
Hugh
Quoted 14 lines · Hugh Dickins
Quoted 11 lines · Hugh Dickins
Quoted 2 lines · Oleg Nesterov
shmem_getpage_gfp(). But, is there any way to figure out that this inode/mapping/aops/whatever is actually shmem?On 3.15 and later, you're in luck: Hannes added bool
shmem_mapping(mapping)in his0cd6144aadd2"mm + fs: prepare for non-page entries in page cache radix trees"; and I just checked, it builds for "tiny" !CONFIG_SHMEM too.If you're backporting to an earlier kernel, it would probably be best to add in a very small patch, extracting just
shmem_mapping()and itslinux/mm.hdeclarations from0cd6144aadd2.
Quoted 6 lines
Looking into something else, I've just been reminded of
mapping_cap_swap_backed(mapping):mm/madvise.chas been using that test for a year. It also returns true onswapper_space, the imaginary mapping you get from a PageSwapCache page: so I suppose it would be wrong to make Hannes'sshmem_mapping()a wrapper to that. But for any backport of your change, it would better than extracting part of0cd6144aadd2.
Yes, this should work, swap_address_space() can't be seen as ->i_mapping,
Quoted 1 line
(Sorry for going on about backporting, when you may have no such intention.)
No, no, thanks for additional info. Perhaps we will actually backport this, at least in rhel. Although in the latter case perhaps it would be better to extract shmem_mapping() from 0cd6144aadd2, it is simple and straightforward.
Thanks,
Oleg.