Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-29 Thread Dumitru Ceara
Hi Ben, If we keep this fix as is can we please have it backported all the way to and including branch-2.10? Right now it's there in master and branch 2.12. ovn-nbctl daemon mode is available since branch-2.10 and might be affected by this leak as it creates new entries in the database. Thanks,

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-22 Thread Damijan Skvarc
Hi Dumitru, The problem is that ovsdb_idl entity is part of library and can be used by different application, where each application instantiates its own parse/unparse callback functions. Library itself does not know how these parse/unparse functions are implemented thus it is not reliable to call

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-19 Thread Dumitru Ceara
Hi Damijan, Ben, Damijan, sorry, I didn't realize you had already reported and fixed the problem in your pending pull request. I wouldn't have sent my patch otherwise. Regarding a single parsed field instead of parsed bits per column I had the impression that it's ok if unparsing is done for all

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-18 Thread Damijan Skvarc
https://mail.openvswitch.org/pipermail/ovs-dev/2019-July/360285.htmlhttps://mail.openvswitch.org/pipermail/ovs-dev/2019-July/360285.html On Thu, 18 Jul 2019, 20:12 Ben Pfaff, wrote: > I guess I missed it. Will you please point it out to me, for example in > the list archive? > > On Thu, Jul 18,

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-18 Thread Ben Pfaff
On Thu, Jul 18, 2019 at 08:52:13AM -0700, Ben Pfaff wrote: > On Wed, Jul 17, 2019 at 09:05:04PM +0200, Dumitru Ceara wrote: > > Once a column is set in a row using ovsdb_idl_txn_write__ we also mark > > the row "parsed". Otherwise the memory allocated by > > sbrec__parse_() will never be freed. Aft

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-18 Thread Ben Pfaff
I guess I missed it. Will you please point it out to me, for example in the list archive? On Thu, Jul 18, 2019 at 06:58:34PM +0200, Damijan Skvarc wrote: > Hmm, the problem of "parsed" flag is that it identifies "all" columns of > certain row have been parsed, however there are CLI tools which mo

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-18 Thread Damijan Skvarc
Hmm, the problem of "parsed" flag is that it identifies "all" columns of certain row have been parsed, however there are CLI tools which modify only individual colums by calling ovsdb_idl_txn_write_() function. In this case and in case parsed flag would be set in ovsdb_idl_txn_write() function then

Re: [ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-18 Thread Ben Pfaff
On Wed, Jul 17, 2019 at 09:05:04PM +0200, Dumitru Ceara wrote: > Once a column is set in a row using ovsdb_idl_txn_write__ we also mark > the row "parsed". Otherwise the memory allocated by > sbrec__parse_() will never be freed. After marking the row > "parsed", the ovsdb_idl_txn_disassemble functi

[ovs-dev] [PATCH] ovsdb-idl: Mark row "parsed" in ovsdb_idl_txn_write__

2019-07-17 Thread Dumitru Ceara
Once a column is set in a row using ovsdb_idl_txn_write__ we also mark the row "parsed". Otherwise the memory allocated by sbrec__parse_() will never be freed. After marking the row "parsed", the ovsdb_idl_txn_disassemble function will properly free the newly allocated memory for the column (ovsdb_