!230 fix CVE-2023-38039

From: @eaglegai 
Reviewed-by: @robertxw 
Signed-off-by: @robertxw

backport-CVE-2023-38039.patch

@@ -0,0 +1,212 @@
+From 3ee79c1674fd6f99e8efca52cd7510e08b766770 Mon Sep 17 00:00:00 2001
+From: Daniel Stenberg <daniel@haxx.se>
+Date: Wed, 2 Aug 2023 23:34:48 +0200
+Subject: [PATCH] http: return error when receiving too large header set
+
+To avoid abuse. The limit is set to 300 KB for the accumulated size of
+all received HTTP headers for a single response. Incomplete research
+suggests that Chrome uses a 256-300 KB limit, while Firefox allows up to
+1MB.
+
+Closes #11582
+---
+ lib/c-hyper.c     | 12 +++++++-----
+ lib/cf-h1-proxy.c |  4 +++-
+ lib/http.c        | 34 ++++++++++++++++++++++++++++++----
+ lib/http.h        |  9 +++++++++
+ lib/pingpong.c    |  4 +++-
+ lib/urldata.h     | 17 ++++++++---------
+ 6 files changed, 60 insertions(+), 20 deletions(-)
+
+diff --git a/lib/c-hyper.c b/lib/c-hyper.c
+index c29983c0b24a6..0b9d9ab478e67 100644
+--- a/lib/c-hyper.c
++++ b/lib/c-hyper.c
+@@ -182,8 +182,11 @@ static int hyper_each_header(void *userdata,
+     }
+   }
+ 
+-  data->info.header_size += (curl_off_t)len;
+-  data->req.headerbytecount += (curl_off_t)len;
++  result = Curl_bump_headersize(data, len, FALSE);
++  if(result) {
++    data->state.hresult = result;
++    return HYPER_ITER_BREAK;
++  }
+   return HYPER_ITER_CONTINUE;
+ }
+ 
+@@ -313,9 +316,8 @@ static CURLcode status_line(struct Curl_easy *data,
+     if(result)
+       return result;
+   }
+-  data->info.header_size += (curl_off_t)len;
+-  data->req.headerbytecount += (curl_off_t)len;
+-  return CURLE_OK;
++  result = Curl_bump_headersize(data, len, FALSE);
++  return result;
+ }
+ 
+ /*
+diff --git a/lib/cf-h1-proxy.c b/lib/cf-h1-proxy.c
+index c9b157c9bccc7..b1d8cb618b7d1 100644
+--- a/lib/cf-h1-proxy.c
++++ b/lib/cf-h1-proxy.c
+@@ -587,7 +587,9 @@ static CURLcode recv_CONNECT_resp(struct Curl_cfilter *cf,
+         return result;
+     }
+ 
+-    data->info.header_size += (long)perline;
++    result = Curl_bump_headersize(data, perline, TRUE);
++    if(result)
++      return result;
+ 
+     /* Newlines are CRLF, so the CR is ignored as the line isn't
+        really terminated until the LF comes. Treat a following CR
+diff --git a/lib/http.c b/lib/http.c
+index f7c71afd7d847..bc78ff97435c4 100644
+--- a/lib/http.c
++++ b/lib/http.c
+@@ -3920,6 +3920,29 @@ static CURLcode verify_header(struct Curl_easy *data)
+   return CURLE_OK;
+ }
+ 
++CURLcode Curl_bump_headersize(struct Curl_easy *data,
++                              size_t delta,
++                              bool connect_only)
++{
++  size_t bad = 0;
++  if(delta < MAX_HTTP_RESP_HEADER_SIZE) {
++    if(!connect_only)
++      data->req.headerbytecount += (unsigned int)delta;
++    data->info.header_size += (unsigned int)delta;
++    if(data->info.header_size > MAX_HTTP_RESP_HEADER_SIZE)
++      bad = data->info.header_size;
++  }
++  else
++    bad = data->info.header_size + delta;
++  if(bad) {
++    failf(data, "Too large response headers: %zu > %zu",
++          bad, MAX_HTTP_RESP_HEADER_SIZE);
++    return CURLE_RECV_ERROR;
++  }
++  return CURLE_OK;
++}
++
++
+ /*
+  * Read any HTTP header lines from the server and pass them to the client app.
+  */
+@@ -4173,8 +4196,9 @@ CURLcode Curl_http_readwrite_headers(struct Curl_easy *data,
+       if(result)
+         return result;
+ 
+-      data->info.header_size += (long)headerlen;
+-      data->req.headerbytecount += (long)headerlen;
++      result = Curl_bump_headersize(data, headerlen, FALSE);
++      if(result)
++        return result;
+ 
+       /*
+        * When all the headers have been parsed, see if we should give
+@@ -4496,8 +4520,10 @@ CURLcode Curl_http_readwrite_headers(struct Curl_easy *data,
+     if(result)
+       return result;
+ 
+-    data->info.header_size += Curl_dyn_len(&data->state.headerb);
+-    data->req.headerbytecount += Curl_dyn_len(&data->state.headerb);
++    result = Curl_bump_headersize(data, Curl_dyn_len(&data->state.headerb),
++                                  FALSE);
++    if(result)
++      return result;
+ 
+     Curl_dyn_reset(&data->state.headerb);
+   }
+diff --git a/lib/http.h b/lib/http.h
+index df3b4e38b8a88..4aeabc345938c 100644
+--- a/lib/http.h
++++ b/lib/http.h
+@@ -64,6 +64,10 @@ extern const struct Curl_handler Curl_handler_wss;
+ 
+ struct dynhds;
+ 
++CURLcode Curl_bump_headersize(struct Curl_easy *data,
++                              size_t delta,
++                              bool connect_only);
++
+ /* Header specific functions */
+ bool Curl_compareheader(const char *headerline,  /* line to check */
+                         const char *header,   /* header keyword _with_ colon */
+@@ -183,6 +187,11 @@ CURLcode Curl_http_auth_act(struct Curl_easy *data);
+ #define EXPECT_100_THRESHOLD (1024*1024)
+ #endif
+ 
++/* MAX_HTTP_RESP_HEADER_SIZE is the maximum size of all response headers
++   combined that libcurl allows for a single HTTP response, any HTTP
++   version. This count includes CONNECT response headers. */
++#define MAX_HTTP_RESP_HEADER_SIZE (300*1024)
++
+ #endif /* CURL_DISABLE_HTTP */
+ 
+ /****************************************************************************
+diff --git a/lib/pingpong.c b/lib/pingpong.c
+index f3f7cb93cb9b7..523bbec189fe6 100644
+--- a/lib/pingpong.c
++++ b/lib/pingpong.c
+@@ -341,7 +341,9 @@ CURLcode Curl_pp_readresp(struct Curl_easy *data,
+       ssize_t clipamount = 0;
+       bool restart = FALSE;
+ 
+-      data->req.headerbytecount += (long)gotbytes;
++      result = Curl_bump_headersize(data, gotbytes, FALSE);
++      if(result)
++        return result;
+ 
+       pp->nread_resp += gotbytes;
+       for(i = 0; i < gotbytes; ptr++, i++) {
+diff --git a/lib/urldata.h b/lib/urldata.h
+index e5446b6840f63..d21aa415dc94b 100644
+--- a/lib/urldata.h
++++ b/lib/urldata.h
+@@ -629,17 +629,16 @@ struct SingleRequest {
+   curl_off_t bytecount;         /* total number of bytes read */
+   curl_off_t writebytecount;    /* number of bytes written */
+ 
+-  curl_off_t headerbytecount;   /* only count received headers */
+-  curl_off_t deductheadercount; /* this amount of bytes doesn't count when we
+-                                   check if anything has been transferred at
+-                                   the end of a connection. We use this
+-                                   counter to make only a 100 reply (without a
+-                                   following second response code) result in a
+-                                   CURLE_GOT_NOTHING error code */
+-
+   curl_off_t pendingheader;      /* this many bytes left to send is actually
+                                     header and not body */
+   struct curltime start;         /* transfer started at this time */
++  unsigned int headerbytecount;  /* only count received headers */
++  unsigned int deductheadercount; /* this amount of bytes doesn't count when
++                                     we check if anything has been transferred
++                                     at the end of a connection. We use this
++                                     counter to make only a 100 reply (without
++                                     a following second response code) result
++                                     in a CURLE_GOT_NOTHING error code */
+   enum {
+     HEADER_NORMAL,              /* no bad header at all */
+     HEADER_PARTHEADER,          /* part of the chunk is a bad header, the rest
+@@ -1089,7 +1088,6 @@ struct PureInfo {
+   int httpversion; /* the http version number X.Y = X*10+Y */
+   time_t filetime; /* If requested, this is might get set. Set to -1 if the
+                       time was unretrievable. */
+-  curl_off_t header_size;  /* size of read header(s) in bytes */
+   curl_off_t request_size; /* the amount of bytes sent in the request(s) */
+   unsigned long proxyauthavail; /* what proxy auth types were announced */
+   unsigned long httpauthavail;  /* what host auth types were announced */
+@@ -1097,6 +1095,7 @@ struct PureInfo {
+   char *contenttype; /* the content type of the object */
+   char *wouldredirect; /* URL this would've been redirected to if asked to */
+   curl_off_t retry_after; /* info from Retry-After: header */
++  unsigned int header_size;  /* size of read header(s) in bytes */
+ 
+   /* PureInfo members 'conn_primary_ip', 'conn_primary_port', 'conn_local_ip'
+      and, 'conn_local_port' are copied over from the connectdata struct in
+

curl.spec

@@ -6,7 +6,7 @@
 
 Name:           curl
 Version:        8.1.2
-Release:        3
+Release:        4
 Summary:        Curl is used in command lines or scripts to transfer data
 License:        curl
 URL:            https://curl.se/
@@ -18,6 +18,7 @@ Patch4:         backport-curl-7.88.0-tests-warnings.patch
 Patch5:         backport-CVE-2023-32001.patch
 Patch6:         backport-vtls-avoid-memory-leak-if-sha256-call-fails.patch
 Patch7:         backport-urlapi-make-sure-zoneid-is-also-duplicated-in-curl_u.patch
+Patch8:         backport-CVE-2023-38039.patch
 
 BuildRequires:  automake brotli-devel coreutils gcc groff krb5-devel
 BuildRequires:  libidn2-devel libnghttp2-devel libpsl-devel
@@ -202,6 +203,12 @@ rm -rf ${RPM_BUILD_ROOT}%{_libdir}/libcurl.la
 %{_mandir}/man3/*
 
 %changelog
+* Thu Sep 14 2023 gaihuiying <eaglegai@163.com> - 8.1.2-4
+- Type:CVE
+- CVE:CVE-2023-38039
+- SUG:NA
+- DESC:fix CVE-2023-38039
+
 * Wed Sep 06 2023 yanglu <yanglu72@h-partners.com> - 8.1.2-3
 - Type:bugfix
 - CVE:NA