| Message ID | 1456738445-876239-1-git-send-email-arnd@arndb.de (mailing list archive) |
|---|---|
| State | Not Applicable |
| Delegated to: | Herbert Xu |
| Headers | show |
On Mon, Feb 29, 2016 at 1:33 AM, Arnd Bergmann <arnd@arndb.de> wrote: > The addition of tracepoints to the page reference tracking had an > unfortunate side-effect in at least one driver that calls put_page > from its exit function, resulting in a link error: > > `.exit.text' referenced in section `__jump_table' of crypto/built-in.o: defined in discarded section `.exit.text' of crypto/built-in.o > > From a cursory look at that this driver, it seems that it may be > doing the wrong thing here anyway, as the page gets allocated > using 'alloc_page()', and should be freed using '__free_page()' > rather than 'put_page()'. > > With this patch, I no longer get any other build errors from the > page_ref patch, so hopefully we can assume that it's always wrong > to call any of those functions from __exit code, and that no other > driver does it. > > Fixes: 0f80830dd044 ("mm/page_ref: add tracepoint to track down page reference manipulation") > Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Dan Williams <dan.j.williams@intel.com> Vinod, will you take this one? -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
2016-03-01 3:04 GMT+09:00 Dan Williams <dan.j.williams@intel.com>: > On Mon, Feb 29, 2016 at 1:33 AM, Arnd Bergmann <arnd@arndb.de> wrote: >> The addition of tracepoints to the page reference tracking had an >> unfortunate side-effect in at least one driver that calls put_page >> from its exit function, resulting in a link error: >> >> `.exit.text' referenced in section `__jump_table' of crypto/built-in.o: defined in discarded section `.exit.text' of crypto/built-in.o >> >> From a cursory look at that this driver, it seems that it may be >> doing the wrong thing here anyway, as the page gets allocated >> using 'alloc_page()', and should be freed using '__free_page()' >> rather than 'put_page()'. >> >> With this patch, I no longer get any other build errors from the >> page_ref patch, so hopefully we can assume that it's always wrong >> to call any of those functions from __exit code, and that no other >> driver does it. >> >> Fixes: 0f80830dd044 ("mm/page_ref: add tracepoint to track down page reference manipulation") >> Signed-off-by: Arnd Bergmann <arnd@arndb.de> > > Acked-by: Dan Williams <dan.j.williams@intel.com> > > Vinod, will you take this one? Problematic patch ("mm/page_ref: ~~~") is not yet merged one. It is on mmotm and this fix should go together with it or before it. I think that handling this fix by Andrew is easier to all. Thanks. -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 01, 2016 at 10:54:50PM +0900, Joonsoo Kim wrote: > 2016-03-01 3:04 GMT+09:00 Dan Williams <dan.j.williams@intel.com>: > > On Mon, Feb 29, 2016 at 1:33 AM, Arnd Bergmann <arnd@arndb.de> wrote: > >> The addition of tracepoints to the page reference tracking had an > >> unfortunate side-effect in at least one driver that calls put_page > >> from its exit function, resulting in a link error: > >> > >> `.exit.text' referenced in section `__jump_table' of crypto/built-in.o: defined in discarded section `.exit.text' of crypto/built-in.o > >> > >> From a cursory look at that this driver, it seems that it may be > >> doing the wrong thing here anyway, as the page gets allocated > >> using 'alloc_page()', and should be freed using '__free_page()' > >> rather than 'put_page()'. > >> > >> With this patch, I no longer get any other build errors from the > >> page_ref patch, so hopefully we can assume that it's always wrong > >> to call any of those functions from __exit code, and that no other > >> driver does it. > >> > >> Fixes: 0f80830dd044 ("mm/page_ref: add tracepoint to track down page reference manipulation") > >> Signed-off-by: Arnd Bergmann <arnd@arndb.de> > > > > Acked-by: Dan Williams <dan.j.williams@intel.com> > > > > Vinod, will you take this one? > > Problematic patch ("mm/page_ref: ~~~") is not yet merged one. It is on mmotm > and this fix should go together with it or before it. I think that > handling this fix by > Andrew is easier to all. Okay fine by me.
On 02/29/2016 10:33 AM, Arnd Bergmann wrote: > The addition of tracepoints to the page reference tracking had an > unfortunate side-effect in at least one driver that calls put_page > from its exit function, resulting in a link error: > > `.exit.text' referenced in section `__jump_table' of crypto/built-in.o: defined in discarded section `.exit.text' of crypto/built-in.o > > From a cursory look at that this driver, it seems that it may be > doing the wrong thing here anyway, as the page gets allocated > using 'alloc_page()', and should be freed using '__free_page()' > rather than 'put_page()'. > > With this patch, I no longer get any other build errors from the > page_ref patch, so hopefully we can assume that it's always wrong > to call any of those functions from __exit code, and that no other > driver does it. Hopefully that's true. If any such driver was leaking references to those pages, so the put_page() didn't actually result in freeing, the explicit __free_page should catch this via built-in checks. > Fixes: 0f80830dd044 ("mm/page_ref: add tracepoint to track down page reference manipulation") Since it's in mmotm which is quilt-based, the commit hash from -next is not stable. > Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> > --- > crypto/async_tx/async_pq.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/crypto/async_tx/async_pq.c b/crypto/async_tx/async_pq.c > index c0748bbd4c08..08b3ac68952b 100644 > --- a/crypto/async_tx/async_pq.c > +++ b/crypto/async_tx/async_pq.c > @@ -444,7 +444,7 @@ static int __init async_pq_init(void) > > static void __exit async_pq_exit(void) > { > - put_page(pq_scribble_page); > + __free_page(pq_scribble_page); > } > > module_init(async_pq_init); > -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/crypto/async_tx/async_pq.c b/crypto/async_tx/async_pq.c index c0748bbd4c08..08b3ac68952b 100644 --- a/crypto/async_tx/async_pq.c +++ b/crypto/async_tx/async_pq.c @@ -444,7 +444,7 @@ static int __init async_pq_init(void) static void __exit async_pq_exit(void) { - put_page(pq_scribble_page); + __free_page(pq_scribble_page); } module_init(async_pq_init);
The addition of tracepoints to the page reference tracking had an unfortunate side-effect in at least one driver that calls put_page from its exit function, resulting in a link error: `.exit.text' referenced in section `__jump_table' of crypto/built-in.o: defined in discarded section `.exit.text' of crypto/built-in.o From a cursory look at that this driver, it seems that it may be doing the wrong thing here anyway, as the page gets allocated using 'alloc_page()', and should be freed using '__free_page()' rather than 'put_page()'. With this patch, I no longer get any other build errors from the page_ref patch, so hopefully we can assume that it's always wrong to call any of those functions from __exit code, and that no other driver does it. Fixes: 0f80830dd044 ("mm/page_ref: add tracepoint to track down page reference manipulation") Signed-off-by: Arnd Bergmann <arnd@arndb.de> --- crypto/async_tx/async_pq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)