public inbox for cygwin-developers@cygwin.com
 help / color / mirror / Atom feed
* [PATCH] Cygwin: cygheap: fix fork error after heap has grown
@ 2023-04-13  2:35 David McFarland
  2023-04-13 12:53 ` David McFarland
  2023-04-13 16:16 ` Corinna Vinschen
  0 siblings, 2 replies; 6+ messages in thread
From: David McFarland @ 2023-04-13  2:35 UTC (permalink / raw)
  To: cygwin-developers; +Cc: David McFarland

2f9b8ff0 introduced a problem where forks would sometimes fail with:

child_copy: cygheap read copy failed, 0x0..0x80044C750, done 0, windows pid 14032, Win32 error 299

When cygheap_max was > CYGHEAP_STORAGE_INITIAL, commit_size would be set to
allocsize(cygheap_max), which is an address, not a size.  VirtualAlloc would be
called to commit commit_size bytes, which would fail, and then child_copy would
be called with zero as the base address.
---
 winsup/cygwin/mm/cygheap.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/winsup/cygwin/mm/cygheap.cc b/winsup/cygwin/mm/cygheap.cc
index 6a20c159a..d614fc7a9 100644
--- a/winsup/cygwin/mm/cygheap.cc
+++ b/winsup/cygwin/mm/cygheap.cc
@@ -87,7 +87,7 @@ cygheap_fixup_in_child (bool execed)
   SIZE_T commit_size = CYGHEAP_STORAGE_INITIAL - CYGHEAP_STORAGE_LOW;
 
   if (child_proc_info->cygheap_max > (void *) CYGHEAP_STORAGE_INITIAL)
-    commit_size = allocsize (child_proc_info->cygheap_max);
+    commit_size = allocsize (child_proc_info->cygheap_max) - CYGHEAP_STORAGE_LOW;
   cygheap = (init_cygheap *) VirtualAlloc ((LPVOID) CYGHEAP_STORAGE_LOW,
 					   CYGHEAP_STORAGE_HIGH
 					   - CYGHEAP_STORAGE_LOW,
-- 
2.39.0.windows.2


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] Cygwin: cygheap: fix fork error after heap has grown
  2023-04-13  2:35 [PATCH] Cygwin: cygheap: fix fork error after heap has grown David McFarland
@ 2023-04-13 12:53 ` David McFarland
  2023-04-13 16:16 ` Corinna Vinschen
  1 sibling, 0 replies; 6+ messages in thread
From: David McFarland @ 2023-04-13 12:53 UTC (permalink / raw)
  To: cygwin-developers

David McFarland <corngood@gmail.com> writes:

> 2f9b8ff0 introduced a problem where forks would sometimes fail with:
>
> child_copy: cygheap read copy failed, 0x0..0x80044C750, done 0, windows pid 14032, Win32 error 299
>
> When cygheap_max was > CYGHEAP_STORAGE_INITIAL, commit_size would be set to
> allocsize(cygheap_max), which is an address, not a size.  VirtualAlloc would be
> called to commit commit_size bytes, which would fail, and then child_copy would
> be called with zero as the base address.
> ---
>  winsup/cygwin/mm/cygheap.cc | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/winsup/cygwin/mm/cygheap.cc b/winsup/cygwin/mm/cygheap.cc
> index 6a20c159a..d614fc7a9 100644
> --- a/winsup/cygwin/mm/cygheap.cc
> +++ b/winsup/cygwin/mm/cygheap.cc
> @@ -87,7 +87,7 @@ cygheap_fixup_in_child (bool execed)
>    SIZE_T commit_size = CYGHEAP_STORAGE_INITIAL - CYGHEAP_STORAGE_LOW;
>  
>    if (child_proc_info->cygheap_max > (void *) CYGHEAP_STORAGE_INITIAL)
> -    commit_size = allocsize (child_proc_info->cygheap_max);
> +    commit_size = allocsize (child_proc_info->cygheap_max) - CYGHEAP_STORAGE_LOW;
>    cygheap = (init_cygheap *) VirtualAlloc ((LPVOID) CYGHEAP_STORAGE_LOW,
>  					   CYGHEAP_STORAGE_HIGH
>  					   - CYGHEAP_STORAGE_LOW,

It might be better to do the subtraction before the call to allocsize,
but this should still work in practice. I'll wait for more feedback
before I post another patch.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] Cygwin: cygheap: fix fork error after heap has grown
  2023-04-13  2:35 [PATCH] Cygwin: cygheap: fix fork error after heap has grown David McFarland
  2023-04-13 12:53 ` David McFarland
@ 2023-04-13 16:16 ` Corinna Vinschen
  2023-04-18  0:25   ` David McFarland
  1 sibling, 1 reply; 6+ messages in thread
From: Corinna Vinschen @ 2023-04-13 16:16 UTC (permalink / raw)
  To: David McFarland; +Cc: cygwin-developers

Hi David,

On Apr 12 23:35, David McFarland wrote:
> 2f9b8ff0 introduced a problem where forks would sometimes fail with:
> 
> child_copy: cygheap read copy failed, 0x0..0x80044C750, done 0, windows pid 14032, Win32 error 299
> 
> When cygheap_max was > CYGHEAP_STORAGE_INITIAL, commit_size would be set to
> allocsize(cygheap_max), which is an address, not a size.  VirtualAlloc would be
> called to commit commit_size bytes, which would fail, and then child_copy would
> be called with zero as the base address.

Can you please append

  Fixes: 2f9b8ff00cce ("Cygwin: decouple cygheap from Cygwin DLL")
  Signed-off-by: David McFarland <your email address>

?

> ---
>  winsup/cygwin/mm/cygheap.cc | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/winsup/cygwin/mm/cygheap.cc b/winsup/cygwin/mm/cygheap.cc
> index 6a20c159a..d614fc7a9 100644
> --- a/winsup/cygwin/mm/cygheap.cc
> +++ b/winsup/cygwin/mm/cygheap.cc
> @@ -87,7 +87,7 @@ cygheap_fixup_in_child (bool execed)
>    SIZE_T commit_size = CYGHEAP_STORAGE_INITIAL - CYGHEAP_STORAGE_LOW;
>  
>    if (child_proc_info->cygheap_max > (void *) CYGHEAP_STORAGE_INITIAL)
> -    commit_size = allocsize (child_proc_info->cygheap_max);
> +    commit_size = allocsize (child_proc_info->cygheap_max) - CYGHEAP_STORAGE_LOW;

Good catch!  Please move the subtraction of CYGHEAP_STORAGE_LOW into the
allocsize call, along the lines of the other usage of allocsize in _csbrk.

Is that what you mean in your followup mail, perhaps?


Thanks!
Corinna

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] Cygwin: cygheap: fix fork error after heap has grown
  2023-04-13 16:16 ` Corinna Vinschen
@ 2023-04-18  0:25   ` David McFarland
  2023-04-18  0:41     ` David McFarland
  0 siblings, 1 reply; 6+ messages in thread
From: David McFarland @ 2023-04-18  0:25 UTC (permalink / raw)
  To: cygwin-developers

Corinna Vinschen <corinna-cygwin@cygwin.com> writes:

> Can you please append
>
>   Fixes: 2f9b8ff00cce ("Cygwin: decouple cygheap from Cygwin DLL")
>   Signed-off-by: David McFarland <your email address>
>
> ?

Sure.

>
> Good catch!  Please move the subtraction of CYGHEAP_STORAGE_LOW into the
> allocsize call, along the lines of the other usage of allocsize in _csbrk.
>
> Is that what you mean in your followup mail, perhaps?

Yeah, exactly.  I'll follow up with a new patch.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH] Cygwin: cygheap: fix fork error after heap has grown
  2023-04-18  0:25   ` David McFarland
@ 2023-04-18  0:41     ` David McFarland
  2023-04-18  8:14       ` Corinna Vinschen
  0 siblings, 1 reply; 6+ messages in thread
From: David McFarland @ 2023-04-18  0:41 UTC (permalink / raw)
  To: cygwin-developers; +Cc: David McFarland

2f9b8ff0 introduced a problem where forks would sometimes fail with:

child_copy: cygheap read copy failed, 0x0..0x80044C750, done 0, windows pid 14032, Win32 error 299

When cygheap_max was > CYGHEAP_STORAGE_INITIAL, commit_size would be set to
allocsize(cygheap_max), which is an address, not a size.  VirtualAlloc would be
called to commit commit_size bytes, which would fail, and then child_copy would
be called with zero as the base address.

Fixes: 2f9b8ff00cce ("Cygwin: decouple cygheap from Cygwin DLL")
Signed-off-by: David McFarland <corngood@gmail.com>
---
 winsup/cygwin/mm/cygheap.cc | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/winsup/cygwin/mm/cygheap.cc b/winsup/cygwin/mm/cygheap.cc
index 6a20c159a..a20ee5972 100644
--- a/winsup/cygwin/mm/cygheap.cc
+++ b/winsup/cygwin/mm/cygheap.cc
@@ -87,7 +87,8 @@ cygheap_fixup_in_child (bool execed)
   SIZE_T commit_size = CYGHEAP_STORAGE_INITIAL - CYGHEAP_STORAGE_LOW;
 
   if (child_proc_info->cygheap_max > (void *) CYGHEAP_STORAGE_INITIAL)
-    commit_size = allocsize (child_proc_info->cygheap_max);
+    commit_size = allocsize ((char *) child_proc_info->cygheap_max
+                   - CYGHEAP_STORAGE_LOW);
   cygheap = (init_cygheap *) VirtualAlloc ((LPVOID) CYGHEAP_STORAGE_LOW,
 					   CYGHEAP_STORAGE_HIGH
 					   - CYGHEAP_STORAGE_LOW,
-- 
2.39.0.windows.2.1.gf04e877035


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] Cygwin: cygheap: fix fork error after heap has grown
  2023-04-18  0:41     ` David McFarland
@ 2023-04-18  8:14       ` Corinna Vinschen
  0 siblings, 0 replies; 6+ messages in thread
From: Corinna Vinschen @ 2023-04-18  8:14 UTC (permalink / raw)
  To: David McFarland; +Cc: cygwin-developers

Hi David,

On Apr 17 21:41, David McFarland wrote:
> 2f9b8ff0 introduced a problem where forks would sometimes fail with:
> 
> child_copy: cygheap read copy failed, 0x0..0x80044C750, done 0, windows pid 14032, Win32 error 299
> 
> When cygheap_max was > CYGHEAP_STORAGE_INITIAL, commit_size would be set to
> allocsize(cygheap_max), which is an address, not a size.  VirtualAlloc would be
> called to commit commit_size bytes, which would fail, and then child_copy would
> be called with zero as the base address.
> 
> Fixes: 2f9b8ff00cce ("Cygwin: decouple cygheap from Cygwin DLL")
> Signed-off-by: David McFarland <corngood@gmail.com>
> ---
>  winsup/cygwin/mm/cygheap.cc | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)

Patch pushed.


Thanks,
Corinna

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2023-04-18  8:14 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-04-13  2:35 [PATCH] Cygwin: cygheap: fix fork error after heap has grown David McFarland
2023-04-13 12:53 ` David McFarland
2023-04-13 16:16 ` Corinna Vinschen
2023-04-18  0:25   ` David McFarland
2023-04-18  0:41     ` David McFarland
2023-04-18  8:14       ` Corinna Vinschen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).