[pacman-dev] [PATCH] Fix FS#8905 (-Sp as non-root) RESUBMIT

[Xavier]

On Sat, Mar 08, 2008 at 06:58:07PM +0100, Nagy Gabor wrote:
> >From c268ac9ee0afa9413ce4c91d4fcc8cb3c91ee0b1 Mon Sep 17 00:00:00 2001
> From: Nagy Gabor <ngaba at bibl.u-szeged.hu>
> Date: Sat, 8 Mar 2008 18:57:26 +0100
> Subject: [PATCH] Fix FS#8905 (-Sp as non-root)
> 
> This patch is quite ugly, but solves the issue without API change.
> 
> Signed-off-by: Nagy Gabor <ngaba at bibl.u-szeged.hu>
> ---
>  lib/libalpm/trans.c |   14 +++++++++++---
>  1 files changed, 11 insertions(+), 3 deletions(-)
> 
> diff --git a/lib/libalpm/trans.c b/lib/libalpm/trans.c
> index 52dbe7a..a19fb28 100644
> --- a/lib/libalpm/trans.c
> +++ b/lib/libalpm/trans.c
> @@ -72,9 +72,11 @@ int SYMEXPORT alpm_trans_init(pmtranstype_t type, pmtransflag_t flags,
>  	ASSERT(handle->trans == NULL, RET_ERR(PM_ERR_TRANS_NOT_NULL, -1));
>  
>  	/* lock db */
> -	handle->lckfd = _alpm_lckmk();
> -	if(handle->lckfd == -1) {
> -		RET_ERR(PM_ERR_HANDLE_LOCK, -1);
> +	if(!(type == PM_TRANS_TYPE_SYNC && flags & PM_TRANS_FLAG_PRINTURIS)) {
> +		handle->lckfd = _alpm_lckmk();
> +		if(handle->lckfd == -1) {
> +			RET_ERR(PM_ERR_HANDLE_LOCK, -1);
> +		}
>  	}
>  
>  	handle->trans = _alpm_trans_new();
> @@ -190,6 +192,7 @@ int SYMEXPORT alpm_trans_interrupt()
>  int SYMEXPORT alpm_trans_release()
>  {
>  	pmtrans_t *trans;
> +	int nolock;
>  
>  	ALPM_LOG_FUNC;
>  
> @@ -200,10 +203,15 @@ int SYMEXPORT alpm_trans_release()
>  	ASSERT(trans != NULL, RET_ERR(PM_ERR_TRANS_NULL, -1));
>  	ASSERT(trans->state != STATE_IDLE, RET_ERR(PM_ERR_TRANS_NULL, -1));
>  
> +	nolock = (trans->type == PM_TRANS_TYPE_SYNC) && (trans->flags & PM_TRANS_FLAG_PRINTURIS);
> +
>  	_alpm_trans_free(trans);
>  	handle->trans = NULL;
>  
>  	/* unlock db */
> +	if(nolock) {
> +		return(0);
> +	}
>  	if(handle->lckfd != -1) {
>  		close(handle->lckfd);
>  		handle->lckfd = -1;
> -- 
> 1.5.3.8
> 
> 

I was thinking about doing something like that as a quick fix for a 3.1.x
release, since some users were not happy.
I didn't test it yet but it looks fine. We agree that this is a ugly hack
though.