[PATCH v2] docs: remove wrong statement about bug in xenstore

[PATCH v2] docs: remove wrong statement about bug in xenstore

[Juergen Gross]

docs/misc/xenstore.txt states that xenstored will use "0" as a valid
transaction id after 2^32 transactions. This is not true. Only
oxenstored has this bug.

Signed-off-by: Juergen Gross <jgross@suse.com>
---
v2: oxenstored has this bug
---
 docs/misc/xenstore.txt | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/docs/misc/xenstore.txt b/docs/misc/xenstore.txt
index c9f4a05..9a02b9d 100644
--- a/docs/misc/xenstore.txt
+++ b/docs/misc/xenstore.txt
@@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
 	tx_id request header field.  When transaction is started whole
 	db is copied; reads and writes happen on the copy.
 	It is not legal to send non-0 tx_id in TRANSACTION_START.
-	Currently xenstored has the bug that after 2^32 transactions
+	Currently oxenstored has the bug that after 2^32 transactions
 	it will allocate the transid 0 for an actual transaction.
+	xenstored (the implementation written in C) doesn't have this
+	bug.
 
 TRANSACTION_END		T|
 TRANSACTION_END		F|
-- 
2.6.6


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [PATCH v2] docs: remove wrong statement about bug in xenstore

[Jan Beulich]

>>> On 24.10.16 at 15:29, <JGross@suse.com> wrote:
> --- a/docs/misc/xenstore.txt
> +++ b/docs/misc/xenstore.txt
> @@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
>  	tx_id request header field.  When transaction is started whole
>  	db is copied; reads and writes happen on the copy.
>  	It is not legal to send non-0 tx_id in TRANSACTION_START.
> -	Currently xenstored has the bug that after 2^32 transactions
> +	Currently oxenstored has the bug that after 2^32 transactions
>  	it will allocate the transid 0 for an actual transaction.

While I know nothing about OCaml, I read Andrew's earlier reply
to mean different behavior would result than the one described,
namely differing between 32-bit and 64-bit builds.

Jan

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [PATCH v2] docs: remove wrong statement about bug in xenstore

[Juergen Gross]

On 24/10/16 15:57, Jan Beulich wrote:
>>>> On 24.10.16 at 15:29, <JGross@suse.com> wrote:
>> --- a/docs/misc/xenstore.txt
>> +++ b/docs/misc/xenstore.txt
>> @@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
>>  	tx_id request header field.  When transaction is started whole
>>  	db is copied; reads and writes happen on the copy.
>>  	It is not legal to send non-0 tx_id in TRANSACTION_START.
>> -	Currently xenstored has the bug that after 2^32 transactions
>> +	Currently oxenstored has the bug that after 2^32 transactions
>>  	it will allocate the transid 0 for an actual transaction.
> 
> While I know nothing about OCaml, I read Andrew's earlier reply
> to mean different behavior would result than the one described,
> namely differing between 32-bit and 64-bit builds.

Do we really still support 32 bit dom0? I thought the support has
ended some time ago?


Juergen

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [PATCH v2] docs: remove wrong statement about bug in xenstore

[Wei Liu]

On Mon, Oct 24, 2016 at 04:43:59PM +0200, Juergen Gross wrote:
> On 24/10/16 15:57, Jan Beulich wrote:
> >>>> On 24.10.16 at 15:29, <JGross@suse.com> wrote:
> >> --- a/docs/misc/xenstore.txt
> >> +++ b/docs/misc/xenstore.txt
> >> @@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
> >>  	tx_id request header field.  When transaction is started whole
> >>  	db is copied; reads and writes happen on the copy.
> >>  	It is not legal to send non-0 tx_id in TRANSACTION_START.
> >> -	Currently xenstored has the bug that after 2^32 transactions
> >> +	Currently oxenstored has the bug that after 2^32 transactions
> >>  	it will allocate the transid 0 for an actual transaction.
> > 
> > While I know nothing about OCaml, I read Andrew's earlier reply
> > to mean different behavior would result than the one described,
> > namely differing between 32-bit and 64-bit builds.
> 
> Do we really still support 32 bit dom0? I thought the support has
> ended some time ago?

Yes, we still support that.

It is the support of 32 bit Xen that has ended.

Wei.

> 
> 
> Juergen

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [PATCH v2] docs: remove wrong statement about bug in xenstore

[Juergen Gross]

On 24/10/16 16:47, Wei Liu wrote:
> On Mon, Oct 24, 2016 at 04:43:59PM +0200, Juergen Gross wrote:
>> On 24/10/16 15:57, Jan Beulich wrote:
>>>>>> On 24.10.16 at 15:29, <JGross@suse.com> wrote:
>>>> --- a/docs/misc/xenstore.txt
>>>> +++ b/docs/misc/xenstore.txt
>>>> @@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
>>>>  	tx_id request header field.  When transaction is started whole
>>>>  	db is copied; reads and writes happen on the copy.
>>>>  	It is not legal to send non-0 tx_id in TRANSACTION_START.
>>>> -	Currently xenstored has the bug that after 2^32 transactions
>>>> +	Currently oxenstored has the bug that after 2^32 transactions
>>>>  	it will allocate the transid 0 for an actual transaction.
>>>
>>> While I know nothing about OCaml, I read Andrew's earlier reply
>>> to mean different behavior would result than the one described,
>>> namely differing between 32-bit and 64-bit builds.
>>
>> Do we really still support 32 bit dom0? I thought the support has
>> ended some time ago?
> 
> Yes, we still support that.
> 
> It is the support of 32 bit Xen that has ended.

Okay, do we need to specify the exact conditions where the bug will
occur, or is v2 of the patch okay, as it is clearly more accurate than
the original text?


Juergen


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [PATCH v2] docs: remove wrong statement about bug in xenstore

[Jan Beulich]

>>> On 24.10.16 at 16:43, <JGross@suse.com> wrote:
> On 24/10/16 15:57, Jan Beulich wrote:
>>>>> On 24.10.16 at 15:29, <JGross@suse.com> wrote:
>>> --- a/docs/misc/xenstore.txt
>>> +++ b/docs/misc/xenstore.txt
>>> @@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
>>>  	tx_id request header field.  When transaction is started whole
>>>  	db is copied; reads and writes happen on the copy.
>>>  	It is not legal to send non-0 tx_id in TRANSACTION_START.
>>> -	Currently xenstored has the bug that after 2^32 transactions
>>> +	Currently oxenstored has the bug that after 2^32 transactions
>>>  	it will allocate the transid 0 for an actual transaction.
>> 
>> While I know nothing about OCaml, I read Andrew's earlier reply
>> to mean different behavior would result than the one described,
>> namely differing between 32-bit and 64-bit builds.
> 
> Do we really still support 32 bit dom0? I thought the support has
> ended some time ago?

I don't recall any statement to that effect.

Jan

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [PATCH v2] docs: remove wrong statement about bug in xenstore

[Wei Liu]

On Mon, Oct 24, 2016 at 04:51:07PM +0200, Juergen Gross wrote:
> On 24/10/16 16:47, Wei Liu wrote:
> > On Mon, Oct 24, 2016 at 04:43:59PM +0200, Juergen Gross wrote:
> >> On 24/10/16 15:57, Jan Beulich wrote:
> >>>>>> On 24.10.16 at 15:29, <JGross@suse.com> wrote:
> >>>> --- a/docs/misc/xenstore.txt
> >>>> +++ b/docs/misc/xenstore.txt
> >>>> @@ -229,8 +229,10 @@ TRANSACTION_START	|			<transid>|
> >>>>  	tx_id request header field.  When transaction is started whole
> >>>>  	db is copied; reads and writes happen on the copy.
> >>>>  	It is not legal to send non-0 tx_id in TRANSACTION_START.
> >>>> -	Currently xenstored has the bug that after 2^32 transactions
> >>>> +	Currently oxenstored has the bug that after 2^32 transactions
> >>>>  	it will allocate the transid 0 for an actual transaction.
> >>>
> >>> While I know nothing about OCaml, I read Andrew's earlier reply
> >>> to mean different behavior would result than the one described,
> >>> namely differing between 32-bit and 64-bit builds.
> >>
> >> Do we really still support 32 bit dom0? I thought the support has
> >> ended some time ago?
> > 
> > Yes, we still support that.
> > 
> > It is the support of 32 bit Xen that has ended.
> 
> Okay, do we need to specify the exact conditions where the bug will
> occur, or is v2 of the patch okay, as it is clearly more accurate than
> the original text?
> 

Having the this patch go in as-is is not accurate.  You probably don't
know the details about ocaml implementations, so I think it would fine
for Andrew to come up with something more accurate.

How about:

Currently oxenstored has the bug that after the internal representation
of transaction id gets overflowed, 0 is returned.

Wei.


> 
> Juergen
> 

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel